]> git.ipfire.org Git - thirdparty/systemd.git/log
thirdparty/systemd.git
8 years agologinctl: print nontrivial properties in logictl show-* 1248/head
Lukas Nykryn [Fri, 11 Sep 2015 11:37:59 +0000 (13:37 +0200)] 
loginctl: print nontrivial properties in logictl show-*

8 years agoMerge pull request #1227 from intelfx/systemctl-legacy-tools-polkit
Lennart Poettering [Thu, 10 Sep 2015 17:53:31 +0000 (19:53 +0200)] 
Merge pull request #1227 from intelfx/systemctl-legacy-tools-polkit

systemctl: fix various aspects of polkit authorization in legacy tools.

8 years agoMerge pull request #1222 from poettering/image-ops-settings
Daniel Mack [Thu, 10 Sep 2015 17:18:18 +0000 (19:18 +0200)] 
Merge pull request #1222 from poettering/image-ops-settings

machine: when removing, renaming, cloning images also care for .nspaw…

8 years agoMerge pull request #1239 from poettering/cgroup-pids
Daniel Mack [Thu, 10 Sep 2015 17:11:29 +0000 (19:11 +0200)] 
Merge pull request #1239 from poettering/cgroup-pids

core: add support for the "pids" cgroup controller

8 years agoMerge pull request #1242 from poettering/no-off_t
Daniel Mack [Thu, 10 Sep 2015 17:01:25 +0000 (19:01 +0200)] 
Merge pull request #1242 from poettering/no-off_t

Drop usage of off_t

8 years agocore: add support for the "pids" cgroup controller 1239/head
Lennart Poettering [Thu, 10 Sep 2015 10:32:16 +0000 (12:32 +0200)] 
core: add support for the "pids" cgroup controller

This adds support for the new "pids" cgroup controller of 4.3 kernels.
It allows accounting the number of tasks in a cgroup and enforcing
limits on it.

This adds two new setting TasksAccounting= and TasksMax= to each unit,
as well as a gloabl option DefaultTasksAccounting=.

This also updated "cgtop" to optionally make use of the new
kernel-provided accounting.

systemctl has been updated to show the number of tasks for each service
if it is available.

This patch also adds correct support for undoing memory limits for units
using a MemoryLimit=infinity syntax. We do the same for TasksMax= now
and hence keep things in sync here.

8 years agoupdate TODO 1242/head
Lennart Poettering [Thu, 10 Sep 2015 16:18:58 +0000 (18:18 +0200)] 
update TODO

8 years agotree-wide: never use the off_t unless glibc makes us use it
Lennart Poettering [Thu, 10 Sep 2015 16:16:18 +0000 (18:16 +0200)] 
tree-wide: never use the off_t unless glibc makes us use it

off_t is a really weird type as it is usually 64bit these days (at least
in sane programs), but could theoretically be 32bit. We don't support
off_t as 32bit builds though, but still constantly deal with safely
converting from off_t to other types and back for no point.

Hence, never use the type anymore. Always use uint64_t instead. This has
various benefits, including that we can expose these values directly as
D-Bus properties, and also that the values parse the same in all cases.

8 years agoMerge pull request #1236 from evverx/systemctl-requisite-of
Lennart Poettering [Thu, 10 Sep 2015 14:40:04 +0000 (16:40 +0200)] 
Merge pull request #1236 from evverx/systemctl-requisite-of

systemctl: add RequisiteOf* as inverses of Requisite and RequisiteOverridable

8 years agoMerge pull request #1238 from again4you/devel/fix_smack_mtab
Lennart Poettering [Thu, 10 Sep 2015 14:36:07 +0000 (16:36 +0200)] 
Merge pull request #1238 from again4you/devel/fix_smack_mtab

smack: label /etc/mtab as "_" when '--with-smack-run-label' is enabled.

8 years agosmack: label /etc/mtab as "_" when '--with-smack-run-label' is enabled. 1238/head
Sangjung Woo [Thu, 10 Sep 2015 12:52:39 +0000 (21:52 +0900)] 
smack: label /etc/mtab as "_" when '--with-smack-run-label' is enabled.

/etc/mtab should be labeled as "_", even though systemd has its own
smack label using '--with-smack-run-label' configuration. This is mainly
because all processes could read that file and the origin of this file
(i.e. /proc/mounts) is labeled as "_". This labels /etc/mtab as "_" when
'--with-smack-run-label' is enabled.

8 years agosystemctl: add RequisiteOf* as inverses of Requisite and RequisiteOverridable 1236/head
Evgeny Vereshchagin [Thu, 10 Sep 2015 11:49:29 +0000 (14:49 +0300)] 
systemctl: add RequisiteOf* as inverses of Requisite and RequisiteOverridable

8 years agoMerge pull request #1226 from poettering/coccinelle-fixes3
Daniel Mack [Thu, 10 Sep 2015 11:46:14 +0000 (13:46 +0200)] 
Merge pull request #1226 from poettering/coccinelle-fixes3

Third round of Coccinelle fixes

8 years agosystemctl: improve clarity of error messages in the logind path 1227/head
Ivan Shapovalov [Wed, 9 Sep 2015 21:41:46 +0000 (00:41 +0300)] 
systemctl: improve clarity of error messages in the logind path

8 years agosystemctl: relax permission checks in halt_main()
Ivan Shapovalov [Wed, 9 Sep 2015 13:06:49 +0000 (16:06 +0300)] 
systemctl: relax permission checks in halt_main()

Thus we allow (non-interactive) polkit auth to kick in for legacy commands
(halt, poweroff, reboot, telinit) as well.

Fixes (another aspect of) issue #213.

8 years agosystemctl: fix logind bus call error handling in halt_main()
Ivan Shapovalov [Wed, 9 Sep 2015 13:04:35 +0000 (16:04 +0300)] 
systemctl: fix logind bus call error handling in halt_main()

Handle -EOPNOTSUPP and -EINPROGRESS like in start_special().

8 years agosystemctl: minor: use IN_SET() instead of repeated equality checks.
Ivan Shapovalov [Wed, 9 Sep 2015 13:01:10 +0000 (16:01 +0300)] 
systemctl: minor: use IN_SET() instead of repeated equality checks.

8 years agoMerge pull request #1234 from medhefgo/master
Lennart Poettering [Thu, 10 Sep 2015 10:10:44 +0000 (12:10 +0200)] 
Merge pull request #1234 from medhefgo/master

cryptsetup-generator: Properly check return code

8 years agocryptsetup-generator: Properly check return code 1234/head
Jan Janssen [Thu, 10 Sep 2015 08:23:49 +0000 (10:23 +0200)] 
cryptsetup-generator: Properly check return code

8 years agoMerge pull request #1230 from swem/master
Martin Pitt [Thu, 10 Sep 2015 04:56:44 +0000 (06:56 +0200)] 
Merge pull request #1230 from swem/master

Add microphone mute keymap for Dell Precision

8 years agoAdd microphone mute keymap for Dell Precision 1230/head
Chen-Han Hsiao (Stanley) [Thu, 10 Sep 2015 03:20:50 +0000 (11:20 +0800)] 
Add microphone mute keymap for Dell Precision

8 years agosystemctl: legacy tools shall never be interactive
Ivan Shapovalov [Mon, 22 Jun 2015 16:04:38 +0000 (19:04 +0300)] 
systemctl: legacy tools shall never be interactive

Fixes (the main concern of) issue #213.

8 years agoMerge pull request #1225 from mbiebl/remove-sphinx-leftover2
Lennart Poettering [Wed, 9 Sep 2015 21:13:58 +0000 (23:13 +0200)] 
Merge pull request #1225 from mbiebl/remove-sphinx-leftover2

README: remove obsolete instructions regarding python-systemd/sphinx

8 years agotree-wide: take benefit of the fact that hashmap_free() returns NULL 1226/head
Lennart Poettering [Wed, 9 Sep 2015 21:12:07 +0000 (23:12 +0200)] 
tree-wide: take benefit of the fact that hashmap_free() returns NULL

And set_free() too.

Another Coccinelle patch.

8 years agoREADME: remove obsolete instructions regarding python-systemd/sphinx 1225/head
Michael Biebl [Wed, 9 Sep 2015 21:10:34 +0000 (23:10 +0200)] 
README: remove obsolete instructions regarding python-systemd/sphinx

8 years agotree-wide: make use of the fact that strv_free() returns NULL
Lennart Poettering [Wed, 9 Sep 2015 21:05:10 +0000 (23:05 +0200)] 
tree-wide: make use of the fact that strv_free() returns NULL

Another Coccinelle patch.

8 years agoMakefile: fix gardel target
Lennart Poettering [Wed, 9 Sep 2015 21:04:40 +0000 (23:04 +0200)] 
Makefile: fix gardel target

8 years agomachine: when removing, renaming, cloning images also care for .nspawn settings file 1222/head
Lennart Poettering [Wed, 9 Sep 2015 16:51:23 +0000 (18:51 +0200)] 
machine: when removing, renaming, cloning images also care for .nspawn settings file

Whenever we remove/rename/clone a machine image, make sure we do the
same for the image's .nspawn settings file.

8 years agoMerge pull request #1220 from zonque/logind-dry-run
Lennart Poettering [Wed, 9 Sep 2015 16:39:01 +0000 (18:39 +0200)] 
Merge pull request #1220 from zonque/logind-dry-run

logind: make dry run command line arguments work again (v2)

8 years agosystemctl: add dry-run support for scheduled shutdowns 1220/head
Daniel Mack [Wed, 9 Sep 2015 15:10:49 +0000 (17:10 +0200)] 
systemctl: add dry-run support for scheduled shutdowns

Prefix the action parameter with "dry-" in case the --dry-run command
line switch was passed.

8 years agologind: allow dry run variants for scheduled shutdowns
Daniel Mack [Wed, 9 Sep 2015 15:05:03 +0000 (17:05 +0200)] 
logind: allow dry run variants for scheduled shutdowns

Allow passing a "dry-" prefix to the action parameter passed to
.ScheduleShutdown(). When strings with this prefix are passed,
the scheduled action will not take place. Instead, an info message
is logged.

8 years agoMerge pull request #1213 from evverx/systemd-notify-log
Lennart Poettering [Wed, 9 Sep 2015 13:54:41 +0000 (15:54 +0200)] 
Merge pull request #1213 from evverx/systemd-notify-log

notify: log error when sd_pid_notify() == 0

8 years agoMerge pull request #1218 from poettering/safe-fclose
Daniel Mack [Wed, 9 Sep 2015 13:45:03 +0000 (15:45 +0200)] 
Merge pull request #1218 from poettering/safe-fclose

util: introduce safe_fclose() and port everything over to it

8 years agoREADME: document new /etc/mtab requirement
Daniel Mack [Wed, 9 Sep 2015 13:27:19 +0000 (15:27 +0200)] 
README: document new /etc/mtab requirement

8 years agoutil: introduce safe_fclose() and port everything over to it 1218/head
Lennart Poettering [Wed, 9 Sep 2015 13:20:10 +0000 (15:20 +0200)] 
util: introduce safe_fclose() and port everything over to it

Adds a coccinelle script to port things over automatically.

8 years agoMerge pull request #1217 from zonque/free-and-replace
Lennart Poettering [Wed, 9 Sep 2015 13:25:15 +0000 (15:25 +0200)] 
Merge pull request #1217 from zonque/free-and-replace

locale: kill free_and_replace()

8 years agolocale: kill free_and_replace() 1217/head
Daniel Mack [Wed, 9 Sep 2015 13:15:14 +0000 (15:15 +0200)] 
locale: kill free_and_replace()

That function really makes little sense, as the open-coded variant
is much more readable. Also, if the 2nd argument is NULL, mfree()
is a much better candidate.

Convert the only users of this function in localed, and then remove it
entirely.

8 years agoMerge pull request #1216 from poettering/coccinelle-fixes-2
Daniel Mack [Wed, 9 Sep 2015 13:12:28 +0000 (15:12 +0200)] 
Merge pull request #1216 from poettering/coccinelle-fixes-2

Coccinelle fixes 2

8 years agotree-wide: replace while(1) by for(;;) everywhere 1216/head
Lennart Poettering [Wed, 9 Sep 2015 12:33:32 +0000 (14:33 +0200)] 
tree-wide: replace while(1) by for(;;) everywhere

Another Coccinelle script.

8 years agotree-wide: update empty-if coccinelle script to cover empty-while and more
Lennart Poettering [Wed, 9 Sep 2015 12:23:02 +0000 (14:23 +0200)] 
tree-wide: update empty-if coccinelle script to cover empty-while and more

Let's also clean up single-line while and for blocks.

8 years agotree-wide: make use of log_error_errno() return value in more cases
Lennart Poettering [Wed, 9 Sep 2015 09:08:23 +0000 (11:08 +0200)] 
tree-wide: make use of log_error_errno() return value in more cases

The previous coccinelle semantic patch that improved usage of
log_error_errno()'s return value, only looked for log_error_errno()
invocations with a single parameter after the error parameter. Update
the patch to handle arbitrary numbers of additional arguments.

8 years agosd-bus: drop weird empty lines
David Herrmann [Wed, 9 Sep 2015 12:57:04 +0000 (14:57 +0200)] 
sd-bus: drop weird empty lines

We should never put empty lines between `if` and `else if`, unless we use
braces.

8 years agoMerge pull request #1215 from maciejaszek/fds_pass_fixes
Lennart Poettering [Wed, 9 Sep 2015 12:44:52 +0000 (14:44 +0200)] 
Merge pull request #1215 from maciejaszek/fds_pass_fixes

sd_pid_notify_with_fds: fix computing msg_controllen

8 years agoMerge pull request #1214 from zonque/mtab
Lennart Poettering [Wed, 9 Sep 2015 12:43:42 +0000 (14:43 +0200)] 
Merge pull request #1214 from zonque/mtab

core: freeze execution if /etc/mtab exists

8 years agosd_pid_notify_with_fds: fix computing msg_controllen 1215/head
Maciej Wereski [Tue, 8 Sep 2015 13:36:30 +0000 (15:36 +0200)] 
sd_pid_notify_with_fds: fix computing msg_controllen

CMSG_SPACE(0) may return value other than 0. This caused sendmsg to fail
with EINVAL, when have_pid or n_fds was 0.

8 years agoMerge pull request #1212 from again4you/devel/fix_smack_label_#3
Lennart Poettering [Wed, 9 Sep 2015 12:01:26 +0000 (14:01 +0200)] 
Merge pull request #1212 from again4you/devel/fix_smack_label_#3

Wrong smack label of the symlink in CGROUP root directory when enabling '--with-smack-run-label' (v3)

8 years agocore: freeze execution if /etc/mtab exists 1214/head
Daniel Mack [Wed, 9 Sep 2015 12:00:23 +0000 (14:00 +0200)] 
core: freeze execution if /etc/mtab exists

The mount monitor that was added to libmount v2.27 requires /etc/mtab to be
non-existant. As systemd now uses that functionality, we cannot monitor any
mounts anymore, and hence not support .mount units.

Systems that have /etc/mtab around as regular file are unsupported by
systemd since a long time.

This patch makes that condition fatal, so we do not boot up with
non-working mount monitor support.

8 years agonotify: log error when sd_pid_notify() == 0 1213/head
Evgeny Vereshchagin [Wed, 9 Sep 2015 11:51:58 +0000 (14:51 +0300)] 
notify: log error when sd_pid_notify() == 0

8 years agosmack: bugfix the smack label of symlink when '--with-smack-run-label' is set 1212/head
Sangjung Woo [Tue, 8 Sep 2015 06:09:40 +0000 (15:09 +0900)] 
smack: bugfix the smack label of symlink when '--with-smack-run-label' is set

Even though systemd has its own smack label since
'--with-smack-run-label' configuration is set, the smack label of each
CGROUP root directory should have the star (i.e. *) label. This is
mainly because current Linux Kernel set the label in this way.
(Refer to smack_d_instantiate() in security/smack/smack_lsm.c)

However, if systemd has its own smack label and arg_join_controllers is
explicitly set or initialized by initialize_join_controllers() function,
current systemd creates the symlink in CGROUP root directory with its
own smack label as below.

lrwxrwxrwx. 1 root root System  11 Dec 31 16:00 cpu -> cpu,cpuacct
dr-xr-xr-x. 4 root root *        0 Dec 31 16:01 cpu,cpuacct
lrwxrwxrwx. 1 root root System  11 Dec 31 16:00 cpuacct -> cpu,cpuacct

This patch fixes that bug by copying the smack label from the origin.

8 years agosmack: introduce new mac_smack_copy() function
Sangjung Woo [Tue, 8 Sep 2015 05:58:22 +0000 (14:58 +0900)] 
smack: introduce new mac_smack_copy() function

This adds a new mac_smack_copy() function in order to read the smack
label from the source and apply it to the destination.

8 years agoimportd: fix typos
Daniel Mack [Wed, 9 Sep 2015 08:10:23 +0000 (10:10 +0200)] 
importd: fix typos

8 years agoMerge pull request #1210 from poettering/import-fixes
Daniel Mack [Wed, 9 Sep 2015 08:08:44 +0000 (10:08 +0200)] 
Merge pull request #1210 from poettering/import-fixes

cgroup fix, nspawn fix, plus change to download .nspawn files in importd

8 years agoMerge pull request #1207 from poettering/coccinelle-fixes
Daniel Mack [Wed, 9 Sep 2015 07:57:29 +0000 (09:57 +0200)] 
Merge pull request #1207 from poettering/coccinelle-fixes

Coccinelle fixes

8 years agonspawn: add missing comma to gperf file 1210/head
Lennart Poettering [Wed, 9 Sep 2015 06:36:20 +0000 (08:36 +0200)] 
nspawn: add missing comma to gperf file

8 years agoMerge pull request #1180 from evverx/dot-from-to-alias-handling
Lennart Poettering [Wed, 9 Sep 2015 06:31:11 +0000 (08:31 +0200)] 
Merge pull request #1180 from evverx/dot-from-to-alias-handling

analyze: add alias handling for --{from,to}-pattern options of the dot command

8 years agotree-wide: drop {} from one-line if blocks 1207/head
Lennart Poettering [Tue, 8 Sep 2015 21:03:38 +0000 (23:03 +0200)] 
tree-wide: drop {} from one-line if blocks

Patch via coccinelle.

8 years agotree-wide: make use of log_error_errno() return value
Lennart Poettering [Tue, 8 Sep 2015 17:30:45 +0000 (19:30 +0200)] 
tree-wide: make use of log_error_errno() return value

Turns this:

        r = -errno;
        log_error_errno(errno, "foo");

into this:

        r = log_error_errno(errno, "foo");

and this:

        r = log_error_errno(errno, "foo");
        return r;

into this:

        return log_error_errno(errno, "foo");

8 years agotree-wide: don't do assignments within if checks
Lennart Poettering [Tue, 8 Sep 2015 17:14:10 +0000 (19:14 +0200)] 
tree-wide: don't do assignments within if checks

Turn this:

       if ((r = foo()) < 0) { ...

into this:

       r = foo();
       if (r < 0) { ...

8 years agotree-wide: drop redundant if checks before safe_close()
Lennart Poettering [Tue, 8 Sep 2015 16:58:28 +0000 (18:58 +0200)] 
tree-wide: drop redundant if checks before safe_close()

Replace this:

        if (fd >= 0)
                safe_close(fd);

by this:

        safe_close(fd);

8 years agotree-wide: make more code use safe_close()
Lennart Poettering [Tue, 8 Sep 2015 16:53:19 +0000 (18:53 +0200)] 
tree-wide: make more code use safe_close()

Replace this:

        close(fd);
        fd = -1;

write this:

        fd = safe_close(fd);

8 years agotree-wide: use coccinelle to patch a lot of code to use mfree()
Lennart Poettering [Tue, 8 Sep 2015 16:43:11 +0000 (18:43 +0200)] 
tree-wide: use coccinelle to patch a lot of code to use mfree()

This replaces this:

        free(p);
        p = NULL;

by this:

        p = mfree(p);

Change generated using coccinelle. Semantic patch is added to the
sources.

8 years agoMerge pull request #1206 from davidmilburn/ata-by-path
Lennart Poettering [Wed, 9 Sep 2015 06:05:53 +0000 (08:05 +0200)] 
Merge pull request #1206 from davidmilburn/ata-by-path

udev: build by-path identifiers for ATA devices.

8 years agoudev: build by-path identifiers for ATA devices. 1206/head
David Milburn [Tue, 8 Sep 2015 20:27:51 +0000 (15:27 -0500)] 
udev: build by-path identifiers for ATA devices.

/dev/disk/by-path

total 0
lrwxrwxrwx. 1 root root  9 Sep  4 10:02 pci-0000:00:1f.2-ata-2 -> ../../sr0
lrwxrwxrwx. 1 root root  9 Sep  4 10:02 pci-0000:00:1f.2-ata-3 -> ../../sdd
lrwxrwxrwx. 1 root root 10 Sep  4 10:02 pci-0000:00:1f.2-ata-3-part1 -> ../../sdd1
lrwxrwxrwx. 1 root root 10 Sep  4 10:02 pci-0000:00:1f.2-ata-3-part2 -> ../../sdd2
lrwxrwxrwx. 1 root root 10 Sep  4 10:02 pci-0000:00:1f.2-ata-3-part3 -> ../../sdd3
lrwxrwxrwx. 1 root root  9 Sep  4 10:02 pci-0000:03:00.0-ata-4 -> ../../sda
lrwxrwxrwx. 1 root root 10 Sep  4 10:02 pci-0000:03:00.0-ata-4-part1 -> ../../sda1
lrwxrwxrwx. 1 root root 10 Sep  4 10:02 pci-0000:03:00.0-ata-4-part2 -> ../../sda2
lrwxrwxrwx. 1 root root  9 Sep  4 10:02 pci-0000:08:00.0-ata-1 -> ../../sdc
lrwxrwxrwx. 1 root root 10 Sep  4 10:02 pci-0000:08:00.0-ata-1-part1 -> ../../sdc1
lrwxrwxrwx. 1 root root 10 Sep  4 10:02 pci-0000:08:00.0-ata-1-part2 -> ../../sdc2

8 years agoMerge pull request #1203 from mus65/fix-esp-options
Lennart Poettering [Tue, 8 Sep 2015 21:20:22 +0000 (23:20 +0200)] 
Merge pull request #1203 from mus65/fix-esp-options

gpt-auto: fix ESP options and description

8 years agogpt-auto: fix ESP options and description 1203/head
Marius Thesing [Tue, 8 Sep 2015 19:25:05 +0000 (21:25 +0200)] 
gpt-auto: fix ESP options and description

Since 59512f21 the parameters were passed in the wrong order, causing the
options to be interpreted as the description.

Also, while "false" was supposed to be passed for "rw", the ESP should actually
be mounted read-write. It just happened to be "true" since the description char*
was passed for "rw".

8 years agoimportd: for .raw and .tar images, try to download .nspawn settings file too
Lennart Poettering [Tue, 8 Sep 2015 16:25:56 +0000 (18:25 +0200)] 
importd: for .raw and .tar images, try to download .nspawn settings file too

8 years agocgroups: make sure the "devices" controller's enum is named the same way as the contr...
Lennart Poettering [Tue, 8 Sep 2015 16:15:50 +0000 (18:15 +0200)] 
cgroups: make sure the "devices" controller's enum is named the same way as the controller in the kernel

Follow-up to 5bf8002a3a6723ce50331c024122078552fb600a.

8 years agoMerge pull request #1190 from poettering/rework-virt
Daniel Mack [Tue, 8 Sep 2015 13:53:56 +0000 (15:53 +0200)] 
Merge pull request #1190 from poettering/rework-virt

basic: rework virtualization detection API

8 years agosd-login: fix sd_seat_get_active() to return ENODATA v226
David Herrmann [Tue, 8 Sep 2015 12:03:22 +0000 (14:03 +0200)] 
sd-login: fix sd_seat_get_active() to return ENODATA

This seems to be an oversight from:
    707b66c66381c899d7ef640e158ffdd5bcff4deb

We have to return ENODATA instead of ENOENT if a requested entry is
non-present. Also fix the call-site in udev to check for these errors.

8 years agobuild: prepare for v226
David Herrmann [Tue, 8 Sep 2015 11:31:57 +0000 (13:31 +0200)] 
build: prepare for v226

Bump version info and update NEWS for the upcoming release.

8 years agoMerge pull request #1201 from torstehu/fix-typo
David Herrmann [Tue, 8 Sep 2015 11:23:24 +0000 (13:23 +0200)] 
Merge pull request #1201 from torstehu/fix-typo

treewide: fix typos

8 years agoMerge pull request #1192 from poettering/sd-bus-container
Daniel Mack [Tue, 8 Sep 2015 10:59:03 +0000 (12:59 +0200)] 
Merge pull request #1192 from poettering/sd-bus-container

sd-bus: pass container bus errors up to calling process

8 years agotreewide: fix typos 1201/head
Torstein Husebø [Tue, 8 Sep 2015 10:23:18 +0000 (12:23 +0200)] 
treewide: fix typos

8 years agoMerge pull request #1196 from evverx/systemctl-add-consists-of
Lennart Poettering [Tue, 8 Sep 2015 09:16:53 +0000 (11:16 +0200)] 
Merge pull request #1196 from evverx/systemctl-add-consists-of

systemctl: add ConsistsOf as the inverse of PartOf

8 years agoMerge pull request #1198 from martinpitt/master
Lennart Poettering [Tue, 8 Sep 2015 09:09:37 +0000 (11:09 +0200)] 
Merge pull request #1198 from martinpitt/master

cgroup-util: fix devices controller

8 years agocgroup-util: fix devices controller 1198/head
Martin Pitt [Tue, 8 Sep 2015 08:42:22 +0000 (10:42 +0200)] 
cgroup-util: fix devices controller

Commit efdb0237 accidentally changed the name of the "devices" cgroup
controller to "device".

8 years agoMerge pull request #1195 from poettering/nspawn-fixes
Daniel Mack [Tue, 8 Sep 2015 08:40:39 +0000 (10:40 +0200)] 
Merge pull request #1195 from poettering/nspawn-fixes

Various nspawn fixes

8 years agosystemctl: add ConsistsOf as the inverse of PartOf 1196/head
Evgeny Vereshchagin [Mon, 7 Sep 2015 23:35:58 +0000 (02:35 +0300)] 
systemctl: add ConsistsOf as the inverse of PartOf

8 years agoupdate TODO
Lennart Poettering [Mon, 7 Sep 2015 23:37:04 +0000 (01:37 +0200)] 
update TODO

8 years agoNEWS: update contributors list
Lennart Poettering [Mon, 7 Sep 2015 23:36:59 +0000 (01:36 +0200)] 
NEWS: update contributors list

8 years agonspawn: also close uid shift socket in the parent 1195/head
Lennart Poettering [Mon, 7 Sep 2015 23:22:46 +0000 (01:22 +0200)] 
nspawn: also close uid shift socket in the parent

We should really close all parent sides of our child/parent socket
pairs.

8 years agonspawn: short reads do not set errno, hence don't try to print it
Lennart Poettering [Mon, 7 Sep 2015 23:22:26 +0000 (01:22 +0200)] 
nspawn: short reads do not set errno, hence don't try to print it

8 years agoinspawn: switch from SOCK_DGRAM to SOCK_SEQPACKET for internal socketpairs
Lennart Poettering [Mon, 7 Sep 2015 23:17:47 +0000 (01:17 +0200)] 
inspawn: switch from SOCK_DGRAM to SOCK_SEQPACKET for internal socketpairs

SOCK_DGRAM and SOCK_SEQPACKET have very similar semantics when used with
socketpair(). However, SOCK_SEQPACKET has the advantage of knowing a
hangup concept, since it is inherently connection-oriented.

Since we use socket pairs to communicate between the nspawn main process
and the nspawn child process, where the child might die abnormally it's
interesting to us to learn about this via hangups if the child side of
the pair is closed. Hence, let's switch to SOCK_SEQPACKET for these
internal communication sockets.

Fixes #956.

8 years agonspawn: properly propagate errors when we fail to set soemthing up
Lennart Poettering [Mon, 7 Sep 2015 23:17:15 +0000 (01:17 +0200)] 
nspawn: properly propagate errors when we fail to set soemthing up

8 years agoMerge pull request #1193 from phomes/typos
Daniel Mack [Mon, 7 Sep 2015 21:16:14 +0000 (23:16 +0200)] 
Merge pull request #1193 from phomes/typos

man: typo fixes

8 years agoman: typo fixes 1193/head
Thomas Hindoe Paaboel Andersen [Mon, 7 Sep 2015 18:06:58 +0000 (20:06 +0200)] 
man: typo fixes

8 years agomachined: improve error message when trying to get a bus in bus-less containers 1192/head
Lennart Poettering [Mon, 7 Sep 2015 17:53:34 +0000 (19:53 +0200)] 
machined: improve error message when trying to get a bus in bus-less containers

Now that we get useful error messages from sd-bus for container
connections, let's make use of this and report better errors back to
machined clients.

Fixes #685.

8 years agosd-bus: when connecting to a container AF_UNIX bus, return error
Lennart Poettering [Mon, 7 Sep 2015 17:52:11 +0000 (19:52 +0200)] 
sd-bus: when connecting to a container AF_UNIX bus, return error

When forking of a child process for connecting to a container, pass
the preicse connection error to the calling process.

We already did this correctly for kdbus busses, let's do so for dbus1
busses, too.

8 years agoMerge pull request #1191 from poettering/nspawn-split
Daniel Mack [Mon, 7 Sep 2015 17:08:39 +0000 (19:08 +0200)] 
Merge pull request #1191 from poettering/nspawn-split

nspawn: split up nspawn.c into multiple smaller .c files

8 years agonspawn: sort and clean up included header list 1191/head
Lennart Poettering [Mon, 7 Sep 2015 16:56:54 +0000 (18:56 +0200)] 
nspawn: sort and clean up included header list

Let's remove unnecessary inclusions, and order the list alphabetically
as suggested in CODING_STYLE now.

8 years agonspawn: remove nspawn.h, it's empty now
Lennart Poettering [Mon, 7 Sep 2015 16:47:34 +0000 (18:47 +0200)] 
nspawn: remove nspawn.h, it's empty now

8 years agonspawn: split out --uid= logic into nspawn-setuid.[ch]
Lennart Poettering [Mon, 7 Sep 2015 16:42:14 +0000 (18:42 +0200)] 
nspawn: split out --uid= logic into nspawn-setuid.[ch]

8 years agonspawn: split out machined registration code to nspawn-register.[ch]
Lennart Poettering [Mon, 7 Sep 2015 16:36:05 +0000 (18:36 +0200)] 
nspawn: split out machined registration code to nspawn-register.[ch]

8 years agonspawn: split out cgroup related calls into nspawn-cgroup.[ch]
Lennart Poettering [Mon, 7 Sep 2015 16:25:04 +0000 (18:25 +0200)] 
nspawn: split out cgroup related calls into nspawn-cgroup.[ch]

8 years agonspawn: split out network related code to nspawn-network.[ch]
Lennart Poettering [Mon, 7 Sep 2015 15:43:45 +0000 (17:43 +0200)] 
nspawn: split out network related code to nspawn-network.[ch]

8 years agonspawn: split all port exposure code into nspawn-expose-port.[ch]
Lennart Poettering [Mon, 7 Sep 2015 14:52:24 +0000 (16:52 +0200)] 
nspawn: split all port exposure code into nspawn-expose-port.[ch]

8 years agonspawn: split out mount related functions into a new nspawn-mount.c file
Lennart Poettering [Mon, 7 Sep 2015 13:59:52 +0000 (15:59 +0200)] 
nspawn: split out mount related functions into a new nspawn-mount.c file

8 years agoMerge pull request #1189 from poettering/unit-start
Daniel Mack [Mon, 7 Sep 2015 12:27:13 +0000 (14:27 +0200)] 
Merge pull request #1189 from poettering/unit-start

unit: move "not supported" check after condition check in unit_start()

8 years agounit: move "not supported" check after condition check in unit_start() 1189/head
Lennart Poettering [Mon, 7 Sep 2015 12:08:24 +0000 (14:08 +0200)] 
unit: move "not supported" check after condition check in unit_start()

Make sure we always check conditions before checking whether the unit
type is supported in unit_start(), since condition checks are "clean
errors", while "not supported" errors are fatal.

This cleans up the boot output of systemd in containers, where a lot of
NOTSUPP lines were shown befor this fix.

This partially reverts 8ff4d2ab0d4758e914aea6d86154d85f2b2c787f which
reorder the checks.

8 years agobasic: rework virtualization detection API 1190/head
Lennart Poettering [Mon, 7 Sep 2015 11:42:47 +0000 (13:42 +0200)] 
basic: rework virtualization detection API

Introduce a proper enum, and don't pass around string ids anymore. This
simplifies things quite a bit, and makes virtualization detection more
similar to architecture detection.