chiark / gitweb /
elogind.git
11 years agonspawn: assume stdout is always writable if it does not support epoll
Michal Schmidt [Fri, 25 Jan 2013 23:16:13 +0000 (00:16 +0100)]
nspawn: assume stdout is always writable if it does not support epoll

stdout can be redirected to a regular file. Regular files don't support epoll.
nspawn failed with: "Failed to register fds in epoll: Operation not permitted".

If stdout does not support epoll, assume it's always writable.

11 years agojob: fix merging with --ignore-dependencies
Michal Schmidt [Fri, 25 Jan 2013 18:54:21 +0000 (19:54 +0100)]
job: fix merging with --ignore-dependencies

This fixes a bug where a job with --ignore-dependencies would wait for
other jobs because it merged into a previously queued job.

11 years agotest: add test for jobs
Michal Schmidt [Fri, 25 Jan 2013 19:10:00 +0000 (20:10 +0100)]
test: add test for jobs

Add a test case for job merging with --ignore-dependencies.

test.sh is copied from TEST-01-*, only lightly modified (this
should be refactored better in the future).
test-jobs.sh is the core of this test.

11 years agotest: factor out testsuite.target, end.service
Michal Schmidt [Fri, 25 Jan 2013 21:19:19 +0000 (22:19 +0100)]
test: factor out testsuite.target, end.service

Tests can use the same testsuite.target.
Add end.service to call poweroff instead of doing it from ExecStopPost
where it may be skipped on failure of ExecStart.

11 years agotest: use non-blocking systemctl calls in testsuite.service
Michal Schmidt [Fri, 25 Jan 2013 18:49:19 +0000 (19:49 +0100)]
test: use non-blocking systemctl calls in testsuite.service

"systemctl poweroff" called from testsuite.service will cause this unit
itself to stop. To avoid deadlock, the call must not be synchronous.

11 years agoreorganize TODO
Lennart Poettering [Fri, 25 Jan 2013 20:11:59 +0000 (21:11 +0100)]
reorganize TODO

11 years agotmpfiles: exclude /tmp/systemd-private-* from cleanup
Zbigniew Jędrzejewski-Szmek [Fri, 25 Jan 2013 16:42:59 +0000 (17:42 +0100)]
tmpfiles: exclude /tmp/systemd-private-* from cleanup

See http://thread.gmane.org/gmane.comp.sysutils.systemd.devel/6874/focus=6891
Should fix https://bugzilla.redhat.com/show_bug.cgi?id=866693

11 years agotmpfiles: introduce type X
Michal Sekletar [Fri, 18 Jan 2013 15:13:08 +0000 (16:13 +0100)]
tmpfiles: introduce type X

Type X will exclude path itself from clean-up. However, if the path is a
directory systemd-tmpfiles will clean-up its content.

In contrast to type x, where path is ignored completely, type X needs some
Age parameter. In order to determine Age parameter, we will look for config
entries of type d or D and pick the best match.  Best match is either
exact match or longest prefix match.

11 years agoAdd _cleanup_pclose_ and fix mismatching pipe close opened by popen()
Zbigniew Jędrzejewski-Szmek [Fri, 25 Jan 2013 15:33:07 +0000 (10:33 -0500)]
Add _cleanup_pclose_ and fix mismatching pipe close opened by popen()

Based-on-patch-by: Thomas Jarosch <thomas.jarosch@intra2net.com>
cppcheck reported:
[src/bootchart/svg.c:791]: (error) Mismatching allocation and deallocation: f

11 years agologind: rework delay inhibition logic
Lennart Poettering [Fri, 25 Jan 2013 05:30:23 +0000 (06:30 +0100)]
logind: rework delay inhibition logic

- Don't allow any locks to be taken while we are in the process of
  executing the specific operation, so that apps are not surprised if a
  suspend/shutdown happens while they rely on their inhibitor.

- Get rid of the Resumed signal, it was a bad idea, and redundant due to
  PrepareForSleep(false), see below.

- Always send out PrepareFor{Shutdown,Sleep} signals, instead of only if
  a delay lock is taken.

- Move PrepareForSleep(false) after we come back from the suspend, so
  that apps can use this as "Resumed" notification. This also has the
  benefit that apps know when to take a new lock.

11 years agologinctl: don't show [UACCESS] info in device tree
Lennart Poettering [Fri, 25 Jan 2013 03:48:34 +0000 (04:48 +0100)]
loginctl: don't show [UACCESS] info in device tree

As the tree doesn't really necessarily show all device node devices and
only those are marked for uaccess it's kinda pointless showing this at
all, since it would give a pretty incomplete impression of the uaccess
information.

11 years agoupdate TODO
Lennart Poettering [Fri, 25 Jan 2013 02:09:41 +0000 (03:09 +0100)]
update TODO

11 years agoupdate TODO
Lennart Poettering [Fri, 25 Jan 2013 02:07:24 +0000 (03:07 +0100)]
update TODO

11 years agoselinux: we don't need that many debug messages by default
Lennart Poettering [Fri, 25 Jan 2013 02:05:08 +0000 (03:05 +0100)]
selinux: we don't need that many debug messages by default

11 years agoselinux-access: Delete debugging message logged as an error
Colin Walters [Thu, 24 Jan 2013 22:47:45 +0000 (17:47 -0500)]
selinux-access: Delete debugging message logged as an error

I don't see why this should be logged at all, so let's delete it.

11 years agoman: mention /proc/self/mountinfo in systemd.mount(5)
Zbigniew Jędrzejewski-Szmek [Fri, 25 Jan 2013 00:59:33 +0000 (19:59 -0500)]
man: mention /proc/self/mountinfo in systemd.mount(5)

11 years agoman: systemd.exec - explicit Environment assignment
Frederic Crozat [Thu, 24 Jan 2013 17:06:00 +0000 (18:06 +0100)]
man: systemd.exec - explicit Environment assignment

Hi all,

while working on another bug, I discovered the "strange" way systemd is
parsing Environment= in .service and thought it was worth documenting
(because I don't expect people to find this syntax by themselves unless
they read the parsing code ;)

Be more verbose about using space in Environment field and not
using value of other variables

Fixes https://bugzilla.redhat.com/show_bug.cgi?id=840260

[zj: expand and reformat the example a bit]

11 years agocore: Set source for mountinfo .mount units
Umut Tezduyar [Wed, 23 Jan 2013 08:23:49 +0000 (09:23 +0100)]
core: Set source for mountinfo .mount units

11 years agoservice: make sure the watchdog timer is not restarted while stopping
Michael Olbrich [Wed, 23 Jan 2013 13:12:16 +0000 (14:12 +0100)]
service: make sure the watchdog timer is not restarted while stopping

A watchdog notification may be handled after the watchdog timer was stopped
while stopping the service. As a result the timer is restarted and the
service may be restarted as well.
The watchdog timestamp is initially set during startup in
service_enter_start_post() and cleared when the timer is stopped. Therefore
it can be used as an indication if the timer should be reset.

11 years agoservice: really stop watchdog timer when stopping
Michael Olbrich [Wed, 23 Jan 2013 13:12:15 +0000 (14:12 +0100)]
service: really stop watchdog timer when stopping

For services without ExecStop= the state SERVICE_STOP is never entered. as
a result the watchdog timer is not stopped and the service is restarted (if
it is configuered to restart).
Stopping the watchdog timer for SERVICE_STOP_SIGTERM as well fixes this.

11 years agobuild-sys: add silent rules for gperf generation
Zbigniew Jędrzejewski-Szmek [Thu, 24 Jan 2013 03:30:12 +0000 (22:30 -0500)]
build-sys: add silent rules for gperf generation

11 years agobuild-sys: add silent rules for xslt processing
Zbigniew Jędrzejewski-Szmek [Thu, 24 Jan 2013 03:26:43 +0000 (22:26 -0500)]
build-sys: add silent rules for xslt processing

11 years agobuild-sys: add silent rule for m4 processing
Zbigniew Jędrzejewski-Szmek [Thu, 24 Jan 2013 03:20:23 +0000 (22:20 -0500)]
build-sys: add silent rule for m4 processing

Using custom prefixes makes the whole build process a bit more
readable.

11 years agobuild-sys: link runlevel targets on install only if sysv compat is enabled
Zbigniew Jędrzejewski-Szmek [Thu, 24 Jan 2013 02:41:30 +0000 (21:41 -0500)]
build-sys: link runlevel targets on install only if sysv compat is enabled

11 years agobuild-sys: add variables to collect unit aliases
Zbigniew Jędrzejewski-Szmek [Wed, 23 Jan 2013 04:51:51 +0000 (23:51 -0500)]
build-sys: add variables to collect unit aliases

11 years agobuild-sys: add a variable to collect .wants links
Zbigniew Jędrzejewski-Szmek [Wed, 23 Jan 2013 03:24:54 +0000 (22:24 -0500)]
build-sys: add a variable to collect .wants links

The idea is to make Makefile.am more declarative and avoid
repetitions. Redeclaring unit links as variables also makes
it easier to conditionally install only some of them.

11 years agohwdb: update
Kay Sievers [Thu, 24 Jan 2013 11:42:57 +0000 (12:42 +0100)]
hwdb: update

11 years agoefi: fix Usec vs. USec
Kay Sievers [Thu, 24 Jan 2013 09:32:21 +0000 (10:32 +0100)]
efi: fix Usec vs. USec

11 years agoREADME: remove mentioning of nss-myhostname "package"
Kay Sievers [Thu, 24 Jan 2013 09:31:34 +0000 (10:31 +0100)]
README: remove mentioning of nss-myhostname "package"

11 years agologinctl: show seat master and uaccess devices in seat-status output
Lennart Poettering [Thu, 24 Jan 2013 04:47:37 +0000 (05:47 +0100)]
loginctl: show seat master and uaccess devices in seat-status output

11 years agoupdate TODO
Lennart Poettering [Thu, 24 Jan 2013 04:33:33 +0000 (05:33 +0100)]
update TODO

11 years agologind: add UnlockSessions() clal to complement LockSessions()
Lennart Poettering [Thu, 24 Jan 2013 04:29:37 +0000 (05:29 +0100)]
logind: add UnlockSessions() clal to complement LockSessions()

11 years agologind: send Resumed() signal after we come back from suspend/hibernate/hybrid-sleep
Lennart Poettering [Thu, 24 Jan 2013 04:15:36 +0000 (05:15 +0100)]
logind: send Resumed() signal after we come back from suspend/hibernate/hybrid-sleep

This allows clients to get asynchronous notifications for user-requested
suspend/hibernate cycles. Kernel-triggered automatic suspending is not
covered.

11 years agologind: only allow one shutdown/sleep action to be queued at the same time
Lennart Poettering [Thu, 24 Jan 2013 03:56:44 +0000 (04:56 +0100)]
logind: only allow one shutdown/sleep action to be queued at the same time

This should make sure that closing the lid while shutting down won't
suspend the machine but will simply cause the shutdown to complete.

11 years agoupdate TODO
Lennart Poettering [Thu, 24 Jan 2013 01:59:25 +0000 (02:59 +0100)]
update TODO

11 years agoReplace autoconf int max test with simple define
Zbigniew Jędrzejewski-Szmek [Wed, 23 Jan 2013 04:05:41 +0000 (23:05 -0500)]
Replace autoconf int max test with simple define

11 years agoefi: read microseconds from boot loader info instead of "ticks" magic
Kay Sievers [Wed, 23 Jan 2013 01:16:11 +0000 (02:16 +0100)]
efi: read microseconds from boot loader info instead of "ticks" magic

11 years agoconfigure.ac: combine --Wformat* options into one argument
Kay Sievers [Tue, 22 Jan 2013 17:10:55 +0000 (18:10 +0100)]
configure.ac: combine --Wformat* options into one argument

https://bugs.freedesktop.org/show_bug.cgi?id=59621

11 years agoman: clearify the meaning of timeout=0 for password agents
Tom Gundersen [Wed, 23 Jan 2013 00:02:14 +0000 (01:02 +0100)]
man: clearify the meaning of timeout=0 for password agents

The fact that timeout=0 makes password agents wait indefinitely is documented
in http://www.freedesktop.org/wiki/Software/systemd/PasswordAgents. Document
it also in the relevant man pages.

11 years agoFix some typos
Eelco Dolstra [Sun, 20 Jan 2013 15:27:58 +0000 (16:27 +0100)]
Fix some typos

11 years agoupdate TODO
Lennart Poettering [Mon, 21 Jan 2013 02:19:36 +0000 (03:19 +0100)]
update TODO

11 years agoupdate TODO
Lennart Poettering [Mon, 21 Jan 2013 00:28:13 +0000 (01:28 +0100)]
update TODO

11 years agoefi: set a nice description string in the ESP units
Lennart Poettering [Mon, 21 Jan 2013 00:27:32 +0000 (01:27 +0100)]
efi: set a nice description string in the ESP units

11 years agoefi: properly create symlink dir
Lennart Poettering [Mon, 21 Jan 2013 00:12:51 +0000 (01:12 +0100)]
efi: properly create symlink dir

11 years agobuild-sys: add makefile to bootchart
Lennart Poettering [Mon, 21 Jan 2013 00:04:05 +0000 (01:04 +0100)]
build-sys: add makefile to bootchart

11 years agoefi: add efi boot generator that automatically mounts the ESP to /boot
Lennart Poettering [Mon, 21 Jan 2013 00:02:53 +0000 (01:02 +0100)]
efi: add efi boot generator that automatically mounts the ESP to /boot

11 years agoutil: move is_efiboot() to efivars.c
Lennart Poettering [Sat, 19 Jan 2013 03:41:33 +0000 (04:41 +0100)]
util: move is_efiboot() to efivars.c

11 years agoshutdown: loop - convert to enumerate match
Kay Sievers [Sun, 20 Jan 2013 15:18:00 +0000 (16:18 +0100)]
shutdown: loop - convert to enumerate match

11 years agoTODO: update loader variables
Kay Sievers [Sun, 20 Jan 2013 03:12:29 +0000 (04:12 +0100)]
TODO: update loader variables

11 years agoexecute: Fix seccomp support on x32
Jonathan Callen [Wed, 25 Jul 2012 02:45:22 +0000 (22:45 -0400)]
execute: Fix seccomp support on x32

In the x32 ABI, syscall numbers start at 0x40000000.  Mask that bit on
x32 for lookups in the syscall_names array and syscall_filter and ensure
that syscall.h is parsed correctly.

[zj: added SYSCALL_TO_INDEX, INDEX_TO_SYSCALL macros.]

11 years agoudev: net_id - add link to the wiki page
Kay Sievers [Sat, 19 Jan 2013 15:01:26 +0000 (16:01 +0100)]
udev: net_id - add link to the wiki page

11 years agocore: if the bootloader supports it, determine firmware and boot loader delay
Lennart Poettering [Sat, 19 Jan 2013 03:20:23 +0000 (04:20 +0100)]
core: if the bootloader supports it, determine firmware and boot loader delay

This allows us to print simple performance data of all parts of the boot now:

- firmware
- boot loader
- kernel
- initrd
- userspace

This only works for bootloaders which support passing TSC data via EFI
variables. As of now that's only gummiboot.

11 years agotmpfiles: do not make /run/nologin executable
Michał Bartoszkiewicz [Fri, 18 Jan 2013 20:33:59 +0000 (21:33 +0100)]
tmpfiles: do not make /run/nologin executable

11 years agoupdate TODO
Lennart Poettering [Sat, 19 Jan 2013 01:11:01 +0000 (02:11 +0100)]
update TODO

11 years agounit: fix typo
Lennart Poettering [Sat, 19 Jan 2013 00:18:01 +0000 (01:18 +0100)]
unit: fix typo

11 years agoupdate TODO
Lennart Poettering [Sat, 19 Jan 2013 00:13:52 +0000 (01:13 +0100)]
update TODO

11 years agoupdate TODO
Lennart Poettering [Thu, 17 Jan 2013 20:20:23 +0000 (21:20 +0100)]
update TODO

11 years agounit: optionally allow making cgroup attribute changes persistent
Lennart Poettering [Sat, 19 Jan 2013 00:01:41 +0000 (01:01 +0100)]
unit: optionally allow making cgroup attribute changes persistent

11 years agocgroup: additional validity checks for cgroup attribute names
Lennart Poettering [Fri, 18 Jan 2013 23:59:19 +0000 (00:59 +0100)]
cgroup: additional validity checks for cgroup attribute names

11 years agoupdate TODO
Lennart Poettering [Fri, 18 Jan 2013 23:11:45 +0000 (00:11 +0100)]
update TODO

11 years agobuild-sys: prune the list of directories we create on install
Lennart Poettering [Fri, 18 Jan 2013 23:11:36 +0000 (00:11 +0100)]
build-sys: prune the list of directories we create on install

11 years agobuild-sys: add autoconf macro to pick macro for x32 compatibility
Zbigniew Jędrzejewski-Szmek [Fri, 18 Jan 2013 21:49:48 +0000 (16:49 -0500)]
build-sys: add autoconf macro to pick macro for x32 compatibility

11 years agoman: update suggested yum command line in nspawn(1)
Lennart Poettering [Fri, 18 Jan 2013 17:22:55 +0000 (18:22 +0100)]
man: update suggested yum command line in nspawn(1)

11 years agonspawn: add audit caps to default set to keep
Lennart Poettering [Fri, 18 Jan 2013 17:13:01 +0000 (18:13 +0100)]
nspawn: add audit caps to default set to keep

Due to the brokeness of much of the userspace audit code we cannot
really start too many systems without the audit caps set. To make nspawn
easier to use just add the audit caps by default.

To boot up containers successfully the kernel's auditing needs to be
turned off still (use "audit=0" on the kernel command line), but at
least no manual caps have to be passed anymore.

In the long run auditing will be fixed for containers and ve virtualized
properly at which time it should be safe to enable these caps anyway.

11 years agocore/cgroup-util: simplify functions and add tests
Zbigniew Jędrzejewski-Szmek [Fri, 18 Jan 2013 06:13:27 +0000 (01:13 -0500)]
core/cgroup-util: simplify functions and add tests

11 years agocore/group-util: merge two functions
Zbigniew Jędrzejewski-Szmek [Fri, 18 Jan 2013 06:13:26 +0000 (01:13 -0500)]
core/group-util: merge two functions

11 years agoman: document the _SYSTEMD_USER_UNIT journal field
Mirco Tischler [Thu, 17 Jan 2013 17:55:09 +0000 (18:55 +0100)]
man: document the _SYSTEMD_USER_UNIT journal field

11 years agosystemctl: print the user session journal for user session units.
Mirco Tischler [Thu, 17 Jan 2013 17:55:08 +0000 (18:55 +0100)]
systemctl: print the user session journal for user session units.

11 years agologs-show: add show_journal_by_user_unit
Mirco Tischler [Thu, 17 Jan 2013 17:55:07 +0000 (18:55 +0100)]
logs-show: add show_journal_by_user_unit

Print the journal for a user session unit. For now this filters by
_SYSTEMD_USER_UNIT and USER_UNIT and additionally _UID.

11 years agocore: log USER_UNIT instead of UNIT if in user session
Mirco Tischler [Thu, 17 Jan 2013 17:55:06 +0000 (18:55 +0100)]
core: log USER_UNIT instead of UNIT if in user session

11 years agojournal: log _SYSTEMD_USER_UNIT for user session units
Mirco Tischler [Thu, 17 Jan 2013 17:55:05 +0000 (18:55 +0100)]
journal: log _SYSTEMD_USER_UNIT for user session units

11 years agoutil: modernization and test for load_env_file
Zbigniew Jędrzejewski-Szmek [Fri, 18 Jan 2013 15:57:11 +0000 (10:57 -0500)]
util: modernization and test for load_env_file

11 years agoutil: continuation support for load_env_file
Michal Vyskocil [Fri, 18 Jan 2013 09:05:10 +0000 (10:05 +0100)]
util: continuation support for load_env_file

Variable definitions can be written on more than one line - if each ends
with a backslash, then is concatenated with a previous one. Only
backslash and unix end of line (\n) are treated as a continuation.

Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=58083
[zj: squashed two patches together; cleaned up grammar; removed
     comment about ignoring trailing backslash -- it is not ignored.]

Document continuation support in systemd.exec

11 years agojournal-gatewayd,man: document new HTTPS options
Zbigniew Jędrzejewski-Szmek [Fri, 18 Jan 2013 06:41:01 +0000 (01:41 -0500)]
journal-gatewayd,man: document new HTTPS options

11 years agojournal-gatewayd: return nice error on unsupported methods
Zbigniew Jędrzejewski-Szmek [Thu, 17 Jan 2013 06:53:01 +0000 (01:53 -0500)]
journal-gatewayd: return nice error on unsupported methods

Returns "HTTP/1.0 406 Not Acceptable" instead of silently
closing the connection.

11 years agojournal-gatewayd: allow pipelining
Zbigniew Jędrzejewski-Szmek [Wed, 28 Nov 2012 17:32:01 +0000 (18:32 +0100)]
journal-gatewayd: allow pipelining

The request must not be answered immediately (at first call to
response_handler()), but on the second. This is also important
for authentication, which cannot be performed on the first call.

Before:

% wget -O/dev/null -S https://localhost:19531/
--2012-11-28 18:29:43--  https://localhost:19531/
Resolving localhost (localhost)... 127.0.0.1
Connecting to localhost (localhost)|127.0.0.1|:19531... connected.
HTTP request sent, awaiting response...
  HTTP/1.1 301 Moved Permanently
  Connection: close
  Content-Length: 87
  Location: /browse
  Content-Type: text/html
  Date: Wed, 28 Nov 2012 17:29:44 GMT
Location: /browse [following]
--2012-11-28 18:29:43--  https://localhost:19531/browse
Connecting to localhost (localhost)|127.0.0.1|:19531... connected.
HTTP request sent, awaiting response...
  HTTP/1.1 200 OK
  Connection: close
  Content-Length: 23260
  Content-Type: text/html
  Date: Wed, 28 Nov 2012 17:29:44 GMT
Length: 23260 (23K) [text/html]

After:

% wget --no-check-certificate -O/dev/null -S https://localhost:19531/
--2012-11-28 18:30:05--  https://localhost:19531/
Resolving localhost (localhost)... 127.0.0.1
Connecting to localhost (localhost)|127.0.0.1|:19531... connected.
HTTP request sent, awaiting response...
  HTTP/1.1 301 Moved Permanently
  Content-Length: 87
  Location: /browse
  Content-Type: text/html
  Date: Wed, 28 Nov 2012 17:30:05 GMT
Location: /browse [following]
--2012-11-28 18:30:05--  https://localhost:19531/browse
Reusing existing connection to localhost:19531.
HTTP request sent, awaiting response...
  HTTP/1.1 200 OK
  Content-Length: 23260
  Content-Type: text/html
  Date: Wed, 28 Nov 2012 17:30:06 GMT
Length: 23260 (23K) [text/html]

11 years agojournal-gatewayd: always log oom() in addition to returning error
Zbigniew Jędrzejewski-Szmek [Wed, 28 Nov 2012 17:00:09 +0000 (18:00 +0100)]
journal-gatewayd: always log oom() in addition to returning error

11 years agoshare/log: skip file/line/func info if empty
Zbigniew Jędrzejewski-Szmek [Mon, 26 Nov 2012 15:39:46 +0000 (16:39 +0100)]
share/log: skip file/line/func info if empty

The new microhttpd logger doesn't know this information. It is
better to log nothing than fake values.

11 years agoshare/log: unify two code paths
Zbigniew Jędrzejewski-Szmek [Mon, 26 Nov 2012 15:39:46 +0000 (16:39 +0100)]
share/log: unify two code paths

Explicit zeroing is replaced with initialization to {0}.

No functional change.

11 years agojournal-gatewayd: redirect microhttpd messages to journal
Zbigniew Jędrzejewski-Szmek [Sun, 25 Nov 2012 22:54:31 +0000 (23:54 +0100)]
journal-gatewayd: redirect microhttpd messages to journal

A prefix ("microhttpd: ") is added to the log lines to make it easy to
distinguish the source.

11 years agojournal-gatewayd: SSL support
Zbigniew Jędrzejewski-Szmek [Sun, 25 Nov 2012 22:26:15 +0000 (23:26 +0100)]
journal-gatewayd: SSL support

For now the certificates are passed around as options to the
program. This might not be the most convenient under "production",
but makes for fairly easy testing.

11 years agojournal-gatewayd: unify two code paths
Zbigniew Jędrzejewski-Szmek [Sun, 25 Nov 2012 22:21:22 +0000 (23:21 +0100)]
journal-gatewayd: unify two code paths

In preparation for adding more options, split out the option
handling code.

11 years agobuild-sys: keep noninstallable tests in noinst_tests
Zbigniew Jędrzejewski-Szmek [Fri, 18 Jan 2013 04:49:05 +0000 (23:49 -0500)]
build-sys: keep noninstallable tests in noinst_tests

Repeating all tests in noinst_PROGRAMS and TESTS is pointless.
This way it is also clearer which noinst_PROGRAMs are not
part of the test suite.

11 years agoupdate TODO
Lennart Poettering [Fri, 18 Jan 2013 00:48:46 +0000 (01:48 +0100)]
update TODO

11 years agosystemctl: add new "get-cgroup-attr" to query current cgroup attribute value
Lennart Poettering [Fri, 18 Jan 2013 00:44:41 +0000 (01:44 +0100)]
systemctl: add new "get-cgroup-attr" to query current cgroup attribute value

Also adds a pair of bus calls for this to the daemon.

11 years agosystemctl: it's probably a good idea not to alter return parameters if we fail
Lennart Poettering [Thu, 17 Jan 2013 23:53:06 +0000 (00:53 +0100)]
systemctl: it's probably a good idea not to alter return parameters if we fail

We generally follow the rule not to touch return values unless we
succeed, so for the sake of uniformity do the same here.

11 years agosystemctl: no need to check this explicitly, we already checked it a few lines up...
Lennart Poettering [Thu, 17 Jan 2013 23:52:45 +0000 (00:52 +0100)]
systemctl: no need to check this explicitly, we already checked it a few lines up anyway, so let's just assert

11 years agosystemctl: we can make this faster and shorten it a bit with strv_extend()
Lennart Poettering [Thu, 17 Jan 2013 23:51:46 +0000 (00:51 +0100)]
systemctl: we can make this faster and shorten it a bit with strv_extend()

Now that strv_extend() is not so slow anymore, we can make use of it, to
shorten our code a bit.

11 years agostrv: make strv_extend() smarter
Lennart Poettering [Thu, 17 Jan 2013 23:47:19 +0000 (00:47 +0100)]
strv: make strv_extend() smarter

11 years agosystemctl: make list-dependencies default to default.target
Lennart Poettering [Thu, 17 Jan 2013 23:36:12 +0000 (00:36 +0100)]
systemctl: make list-dependencies default to default.target

11 years agosystemctl: we can use nulstr_contains() for this lookup
Lennart Poettering [Thu, 17 Jan 2013 23:29:47 +0000 (00:29 +0100)]
systemctl: we can use nulstr_contains() for this lookup

It's a bit easier to read...

11 years agohwdb: update
Kay Sievers [Thu, 17 Jan 2013 23:15:25 +0000 (00:15 +0100)]
hwdb: update

11 years agosystemctl add command list-dependencies
Lukas Nykryn [Thu, 17 Jan 2013 20:34:11 +0000 (21:34 +0100)]
systemctl add command list-dependencies

systemctl list-dependencies lists all unit's dependecies and
recursively expands all subsidiary target units into a tree.

Primary purpose for this command is to show all units which are
enabled in specified target.

11 years agoutil: fix bad memory access
Lennart Poettering [Thu, 17 Jan 2013 16:38:00 +0000 (17:38 +0100)]
util: fix bad memory access

11 years agoTODO: remove vconsole items
Kay Sievers [Thu, 17 Jan 2013 16:00:50 +0000 (17:00 +0100)]
TODO: remove vconsole items

We should not pimp up the kernel's VC stuff, it's too linited and
fragile.

At the moment not even the font uploaded early during bootup does
survive the KMS driver taking over the framebuffer driver. We
surely don't want to make promises about colors or resolution.

The future is fullscreen KMS/kmscon/wayland/... based terminals using
X fonts, keymaps, input methods, and not the old school too limited
kernel VC stuff. So leave the kernel VCs as they are, and don't expect
wonders.

11 years agocore: corrects check of strduped controller string
Nestor Ovroy [Thu, 17 Jan 2013 13:55:30 +0000 (05:55 -0800)]
core: corrects check of strduped controller string

In commit 246aa6d (core: add bus API and systemctl commands for altering
cgroup parameters during runtime), when rewriting unit_add_one_default_cgroup
to prefered style, the check of strduped b->controller was incorrectly
changed to check the containing structure. Correct it.

11 years agoservice: properly signal permanent failure of a service to its socket
Lennart Poettering [Thu, 17 Jan 2013 03:52:19 +0000 (04:52 +0100)]
service: properly signal permanent failure of a service to its socket

This makes sure that a service is not indefinitely restarted in a tight
loop if it fails before it is able to process its socket.

This corrects the breakage introduced with
8d1b002a2e389e79a2414491523de549783abf73. Shame on me.

11 years agoTODO: update udev
Kay Sievers [Thu, 17 Jan 2013 02:30:55 +0000 (03:30 +0100)]
TODO: update udev

11 years agoudev: net_id - suppress bcma core == 0
Kay Sievers [Thu, 17 Jan 2013 02:30:07 +0000 (03:30 +0100)]
udev: net_id - suppress bcma core == 0

11 years agoupdate TODO
Lennart Poettering [Thu, 17 Jan 2013 01:49:57 +0000 (02:49 +0100)]
update TODO