chiark / gitweb /
8 years agojournalctl: for now complain if more than one match is provided since this is still...
Lennart Poettering [Wed, 30 May 2012 20:30:35 +0000 (22:30 +0200)]
journalctl: for now complain if more than one match is provided since this is still broken

8 years agobuild-sys: fix built with --disable-logind
Lennart Poettering [Wed, 30 May 2012 20:25:01 +0000 (22:25 +0200)]
build-sys: fix built with --disable-logind

8 years agodbus: better don't expose options we better shouldn't exposed
Lennart Poettering [Wed, 30 May 2012 19:43:11 +0000 (21:43 +0200)]
dbus: better don't expose options we better shouldn't exposed

NotifySocket= is hardly useful, people should use $NOTIFY_SOCKET if

RunningAs= suggests people had a choice of running user mode as PID 1 or
vice versa, so remove this too.

8 years agoservice: mark compat options as such
Lennart Poettering [Wed, 30 May 2012 19:42:07 +0000 (21:42 +0200)]
service: mark compat options as such

This moves FsckPassNo= and SysVStartPriority= into its own
"Compatibility Options" section in the man page to clarify that these
options are not useful for anything but establishing a limited amount of

Also stop exposing these options on the bus.

8 years agologind: rework button setting semantics
Lennart Poettering [Wed, 30 May 2012 19:40:32 +0000 (21:40 +0200)]
logind: rework button setting semantics

If a graphical session without full DE that handles power/suspend events
is used this can now be controlled by logind instead, optionally.

8 years agologin: Use the GOTO properly in udev rule
Tero Roponen [Wed, 30 May 2012 16:34:28 +0000 (18:34 +0200)]
login: Use the GOTO properly in udev rule

In udev rules the labels must be assigned to GOTO keys with '='.

8 years agojournald: ignore messages read from /proc/kmsg that we generated via /dev/kmsg
Lennart Poettering [Wed, 30 May 2012 13:42:35 +0000 (15:42 +0200)]
journald: ignore messages read from /proc/kmsg that we generated via /dev/kmsg

Avoid a busy loop.

8 years agologind: add missing files
Lennart Poettering [Wed, 30 May 2012 13:07:19 +0000 (15:07 +0200)]
logind: add missing files

8 years agosleep: print nice messages right before and right after system hibernate/sleep
Lennart Poettering [Wed, 30 May 2012 13:04:39 +0000 (15:04 +0200)]
sleep: print nice messages right before and right after system hibernate/sleep

8 years agologind: optionally handle power, sleep and lid switch events
Lennart Poettering [Wed, 30 May 2012 13:01:51 +0000 (15:01 +0200)]
logind: optionally handle power, sleep and lid switch events

This takes handling of chassis power and sleep keys as well as the lid
switch over from acpid.

This logic is enabled by default for power and sleep keys, but not for
the lid switch.

If a graphical session is in the foreground no action is taken under the
assumption that the graphical session does this.

8 years agocapabilities: when dropping capabilities system-wide also drop them from usermode...
Lennart Poettering [Tue, 29 May 2012 21:33:38 +0000 (23:33 +0200)]
capabilities: when dropping capabilities system-wide also drop them from usermode helpers

This hooks things up with /proc/sys/kernel/usermodehelper/bset and

8 years agoupdate TODO
Kay Sievers [Tue, 29 May 2012 21:15:34 +0000 (23:15 +0200)]
update TODO

8 years agoplymouth: use systemd-udev-settle.service
Kay Sievers [Tue, 29 May 2012 20:32:41 +0000 (22:32 +0200)]
plymouth: use systemd-udev-settle.service

8 years agocgtop: work even if not all cgroups are available
Shawn Landden [Tue, 22 May 2012 05:54:41 +0000 (22:54 -0700)]
cgtop: work even if not all cgroups are available

cgtop quits on startup if all the cgroup mounts it expects are not available.
Just continue without nonexistant ones.

8 years agobash: Reflect new name of loginctl in bash-completion script
Gert Michael Kulyk [Sun, 27 May 2012 18:43:55 +0000 (20:43 +0200)]
bash: Reflect new name of loginctl in bash-completion script

8 years agoTODO
Lennart Poettering [Fri, 25 May 2012 12:38:48 +0000 (14:38 +0200)]

8 years agoupdate TODO
Kay Sievers [Tue, 29 May 2012 10:19:40 +0000 (12:19 +0200)]
update TODO

8 years agoman/systemctl.xml: Add missing space for »stopcommand«
Paul Menzel [Thu, 24 May 2012 20:20:39 +0000 (22:20 +0200)]
man/systemctl.xml: Add missing space for »stopcommand«

8 years agotest/README.testsuite: add documentation for the testsuite
Harald Hoyer [Fri, 25 May 2012 16:44:42 +0000 (18:44 +0200)]
test/README.testsuite: add documentation for the testsuite

8 years agotest/TEST-01-BASIC: add /var on cryptoluks and wait for the journal
Harald Hoyer [Fri, 25 May 2012 16:32:55 +0000 (18:32 +0200)]
test/TEST-01-BASIC: add /var on cryptoluks and wait for the journal

8 years agotest/test-functions: add check_qemu() and check_nspawn() functions
Harald Hoyer [Fri, 25 May 2012 16:31:58 +0000 (18:31 +0200)]
test/test-functions: add check_qemu() and check_nspawn() functions

8 years agotest/TEST-01-BASIC: fail test, if qemu failed
Harald Hoyer [Fri, 25 May 2012 12:55:26 +0000 (14:55 +0200)]
test/TEST-01-BASIC: fail test, if qemu failed

8 years agotest/TEST-01-BASIC: make DEBUGTOOLS an environment variable
Harald Hoyer [Fri, 25 May 2012 12:54:38 +0000 (14:54 +0200)]
test/TEST-01-BASIC: make DEBUGTOOLS an environment variable

8 years agotest/test-functions: be more verbose about skipping tests
Harald Hoyer [Fri, 25 May 2012 12:53:29 +0000 (14:53 +0200)]
test/test-functions: be more verbose about skipping tests

8 years agoupdate TODO
Kay Sievers [Tue, 29 May 2012 00:08:59 +0000 (02:08 +0200)]
update TODO

8 years agoudev: man - mention systemd-udevd(8)
Kay Sievers [Mon, 28 May 2012 22:53:28 +0000 (00:53 +0200)]
udev: man - mention systemd-udevd(8)

8 years agoudev.pc: install udev files to /lib/udev rather than /lib/systemd
Tom Gundersen [Sat, 26 May 2012 23:24:16 +0000 (01:24 +0200)]
udev.pc: install udev files to /lib/udev rather than /lib/systemd

8 years agojournald: refer to the correct man page
Matthias Clasen [Sat, 26 May 2012 21:37:39 +0000 (17:37 -0400)]
journald: refer to the correct man page

Fix the reference to the journald.conf man page in the
config file of the same name.

8 years agorules: Limit USB autosuspend on USB HID devices
Matthew Garrett [Fri, 25 May 2012 14:27:25 +0000 (10:27 -0400)]
rules: Limit USB autosuspend on USB HID devices

Some USB ports on external hubs may be reported as "fixed". We only want
to auto-enable this on ports that are internal to the machine, so check
the parent state as well.

8 years agoscsi_id: fix truncated model in by-id device links
Jean Delvare [Fri, 25 May 2012 14:08:00 +0000 (16:08 +0200)]
scsi_id: fix truncated model in by-id device links

Do not forget the leading "S" when appending the serial number,
otherwise we chop the last character of the model name.

Signed-off-by: Jean Delvare <>
Signed-off-by: Petr Uzel <>
8 years agoudev: silent gcc warnings
Kay Sievers [Fri, 25 May 2012 12:25:46 +0000 (14:25 +0200)]
udev: silent gcc warnings

8 years agoNEWS: add contributors list
Lennart Poettering [Thu, 24 May 2012 16:15:35 +0000 (18:15 +0200)]
NEWS: add contributors list

8 years agoswap: avoid bad memory access v183
Lennart Poettering [Thu, 24 May 2012 15:33:50 +0000 (17:33 +0200)]
swap: avoid bad memory access

8 years agoNEWS: fix even more typos
Lennart Poettering [Thu, 24 May 2012 15:06:03 +0000 (17:06 +0200)]
NEWS: fix even more typos

8 years agoupdate TODO
Lennart Poettering [Thu, 24 May 2012 15:04:04 +0000 (17:04 +0200)]
update TODO

8 years agoNEWS: fix more typos
Lennart Poettering [Thu, 24 May 2012 15:03:52 +0000 (17:03 +0200)]
NEWS: fix more typos

8 years agoNEWS: fix more typos
Lennart Poettering [Thu, 24 May 2012 14:53:31 +0000 (16:53 +0200)]
NEWS: fix more typos

8 years agoNEWS: fixo
Lennart Poettering [Thu, 24 May 2012 14:52:12 +0000 (16:52 +0200)]
NEWS: fixo

8 years agoNEWS: explain the version jump
Lennart Poettering [Thu, 24 May 2012 14:41:08 +0000 (16:41 +0200)]
NEWS: explain the version jump

8 years agoupdate TODO
Lennart Poettering [Thu, 24 May 2012 14:39:55 +0000 (16:39 +0200)]
update TODO

8 years agotest/TEST-01-BASIC: add missing "; then"
Harald Hoyer [Thu, 24 May 2012 14:07:35 +0000 (16:07 +0200)]
test/TEST-01-BASIC: add missing "; then"

8 years agotest/TEST-01-BASIC: only test systemd-nspawn, if the test system uses systemd
Harald Hoyer [Thu, 24 May 2012 14:00:33 +0000 (16:00 +0200)]
test/TEST-01-BASIC: only test systemd-nspawn, if the test system uses systemd

8 years agotest/TEST-01-BASIC: add systemd-nspawn run with the root image
Harald Hoyer [Thu, 24 May 2012 13:51:10 +0000 (15:51 +0200)]
test/TEST-01-BASIC: add systemd-nspawn run with the root image

8 years agobuild-sys: prepare release 183
Lennart Poettering [Thu, 24 May 2012 13:42:21 +0000 (15:42 +0200)]
build-sys: prepare release 183

8 years agotest/TEST-01-BASIC: untabified
Harald Hoyer [Thu, 24 May 2012 13:27:53 +0000 (15:27 +0200)]
test/TEST-01-BASIC: untabified

8 years agotest/TEST-01-BASIC: add a defined rc.local
Harald Hoyer [Thu, 24 May 2012 13:27:13 +0000 (15:27 +0200)]
test/TEST-01-BASIC: add a defined rc.local

8 years agotest/TEST-01-BASIC: add /etc/localtime to test image
Harald Hoyer [Thu, 24 May 2012 13:26:09 +0000 (15:26 +0200)]
test/TEST-01-BASIC: add /etc/localtime to test image

8 years agotest: add touch to test machine
Lennart Poettering [Thu, 24 May 2012 13:21:43 +0000 (15:21 +0200)]
test: add touch to test machine

8 years agotest: check systemctl --failed in test VM
Lennart Poettering [Thu, 24 May 2012 12:32:48 +0000 (14:32 +0200)]
test: check systemctl --failed in test VM

8 years agomain: add configuration option to alter capability bounding set for PID 1
Lennart Poettering [Thu, 24 May 2012 02:00:56 +0000 (04:00 +0200)]
main: add configuration option to alter capability bounding set for PID 1

This also ensures that caps dropped from the bounding set are also
dropped from the inheritable set, to be extra-secure. Usually that should
change very little though as the inheritable set is empty for all our uses

8 years agoservice: for Type=idle units consider START_PRE, START, START_POST all as ACTIVE
Lennart Poettering [Thu, 24 May 2012 00:22:35 +0000 (02:22 +0200)]
service: for Type=idle units consider START_PRE, START, START_POST all as ACTIVE

We want to avoid a deadlock when a service has ExecStartPre= programs
that wait for the job queue to run empty because of Type=idle, but which
themselves keep the queue non-empty because START_PRE was considered
ACTIVATING and hence the job not complete. With this patch we alter the
state translation table so that it is impossible ever to wait for
Type=idle unit, hence removing the deadlock.

8 years agoupdate TODO
Lennart Poettering [Wed, 23 May 2012 10:52:39 +0000 (12:52 +0200)]
update TODO

8 years agounits: fix file syntax
Lennart Poettering [Wed, 23 May 2012 10:49:49 +0000 (12:49 +0200)]
units: fix file syntax

8 years agojournal: don't complain if SELinux userspace is available but the kernel lacks it
Lennart Poettering [Wed, 23 May 2012 10:49:41 +0000 (12:49 +0200)]
journal: don't complain if SELinux userspace is available but the kernel lacks it

8 years agoman: properly document FsckPassNo= for mount units
Lennart Poettering [Wed, 23 May 2012 02:11:29 +0000 (04:11 +0200)]
man: properly document FsckPassNo= for mount units

8 years agoreadahead: use Conflicts= instead of masking to get rid of readahead units
Lennart Poettering [Wed, 23 May 2012 02:02:56 +0000 (04:02 +0200)]
readahead: use Conflicts= instead of masking to get rid of readahead units

8 years agoupdate TODO
Lennart Poettering [Wed, 23 May 2012 01:51:55 +0000 (03:51 +0200)]
update TODO

8 years agoreadahead: disable collect/replay during system updates
Lennart Poettering [Wed, 23 May 2012 01:51:48 +0000 (03:51 +0200)]
readahead: disable collect/replay during system updates

8 years agoreadahead: avoid running of readahead services if readahead is already done
Lennart Poettering [Wed, 23 May 2012 01:51:25 +0000 (03:51 +0200)]
readahead: avoid running of readahead services if readahead is already done

8 years agoreadahead: avoid activating the journal by accident, log directly to kmsg
Lennart Poettering [Wed, 23 May 2012 01:50:53 +0000 (03:50 +0200)]
readahead: avoid activating the journal by accident, log directly to kmsg

8 years agoreadahead: when doing a system update drop readahead file
Lennart Poettering [Wed, 23 May 2012 01:50:15 +0000 (03:50 +0200)]
readahead: when doing a system update drop readahead file

8 years agomanager: rework generator logic
Lennart Poettering [Wed, 23 May 2012 01:43:29 +0000 (03:43 +0200)]
manager: rework generator logic

Previously generated units were always placed at the end of the search
path. With this change there will be three unit dirs instead of one, to
place generated entries at the beginning, in the middle and at the end
of the search path:

beginning: for units that need to override all configuration, regardless
of user or vendor. Example use: system-update-generator uses this to
temporarily redirect

middle: for units that need to override vendor configuration, but not
vendor configuration. Example use: /etc/fstab should override vendor
supplied configuration (think /tmp), but should not override native user

end: does not override anything but is available as well. Possible usage
might be to convert D-Bus bus service files to native units but allowing
vendor supplied native units to win.

8 years agoupdate TODO
Lennart Poettering [Tue, 22 May 2012 21:11:21 +0000 (23:11 +0200)]
update TODO

8 years agounits: remove service sysv_path variable and replace it by generic unit_path
Lennart Poettering [Tue, 22 May 2012 21:08:24 +0000 (23:08 +0200)]
units: remove service sysv_path variable and replace it by generic unit_path

UnitPath= is also writable via native units and may be used by generators
to clarify from which file a unit is generated. This patch also hooks up
the cryptsetup and fstab generators to set UnitPath= accordingly.

8 years agolog: make sure generators never log into the journal to avoid activation deadlocks
Lennart Poettering [Tue, 22 May 2012 20:00:37 +0000 (22:00 +0200)]
log: make sure generators never log into the journal to avoid activation deadlocks

This makes all generators log to kmsg by default.

8 years agomount: drop unused function
Lennart Poettering [Tue, 22 May 2012 17:50:10 +0000 (19:50 +0200)]
mount: drop unused function

8 years agosystem-update: add system update generator
Lennart Poettering [Tue, 22 May 2012 17:48:51 +0000 (19:48 +0200)]
system-update: add system update generator

8 years agojournal: log journal internal messages to kmsg
Lennart Poettering [Tue, 22 May 2012 17:26:51 +0000 (19:26 +0200)]
journal: log journal internal messages to kmsg

8 years agoexecute: use a much lower idle timeout that default time
Lennart Poettering [Tue, 22 May 2012 17:26:13 +0000 (19:26 +0200)]
execute: use a much lower idle timeout that default time

The idle timeout after all is for cosmetics only, hence avoid any
substantial delays just for it.

8 years agomount: replace PID1 internal fstab parser with generator
Lennart Poettering [Tue, 22 May 2012 17:23:33 +0000 (19:23 +0200)]
mount: replace PID1 internal fstab parser with generator

Bit by bit we should remove non-unit parsing from PID 1 and move into
generators, to clean up our code base a bit and clearly separate

8 years agomain: only fall back to /bin/sh in case /sbin/init does not exist
Kay Sievers [Tue, 22 May 2012 17:11:10 +0000 (19:11 +0200)]
main: only fall back to /bin/sh in case /sbin/init does not exist

8 years agotest: introduce a basic testsuite framework
Harald Hoyer [Mon, 21 May 2012 16:55:48 +0000 (18:55 +0200)]
test: introduce a basic testsuite framework

$ cd test
$ sudo make check

will run all tests in the TEST-* subdirectories

$ cd test/TEST-01-BASIC
$ sudo make clean setup run

will run the different stages of the test for debugging purposes

8 years agologind: fix write out of user state file
Lennart Poettering [Tue, 22 May 2012 14:46:11 +0000 (16:46 +0200)]
logind: fix write out of user state file

8 years agoswitch-root: do not use close old_root_fd after rm_rf_children()
Harald Hoyer [Tue, 22 May 2012 13:28:45 +0000 (15:28 +0200)]
switch-root: do not use close old_root_fd after rm_rf_children()

rm_rf_children() has already closed the fd with closedir().

8 years agoutil: make sure to fstatat() at most once in rm_rf_children()
Lennart Poettering [Tue, 22 May 2012 14:14:34 +0000 (16:14 +0200)]
util: make sure to fstatat() at most once in rm_rf_children()

8 years agorescue: don't pull in sockets
Lennart Poettering [Tue, 22 May 2012 14:12:25 +0000 (16:12 +0200)]
rescue: don't pull in sockets

In rescue mode let's not establish all sockets, so that we don't end up
starting a lot of additional services automatically.

Instead of pulling in we now only pull in
which pulls in and That way rescue mode has
all the really basic setup around, but normal services are not started
and not autostarted either.

8 years agojournal: crash when filesystem is low on space
Sjoerd Simons [Sat, 14 Apr 2012 12:11:08 +0000 (14:11 +0200)]
journal: crash when filesystem is low on space

When space is getting too low on a file system rotating the journal file
will fail after the rotation, as opening the new logfile will fail.

Recognize this when logging the error and don't try to dereference a
NULL JournalFile pointer.

8 years agomain: properly queue after switched root
Lennart Poettering [Tue, 22 May 2012 00:35:22 +0000 (02:35 +0200)]
main: properly queue after switched root

This also introduces rd.systemd.unit= to specify the unit to boot into
in the initrd.

8 years agobuild-sys: use libpci's pkgconfig file to find the pci.ids file
Kay Sievers [Tue, 22 May 2012 00:35:44 +0000 (02:35 +0200)]
build-sys: use libpci's pkgconfig file to find the pci.ids file

8 years agokeymap: support for Lenovo IdeaPad Z570 keyboard
Maxim A. Mikityanskiy [Tue, 22 May 2012 00:11:21 +0000 (02:11 +0200)]
keymap: support for Lenovo IdeaPad Z570 keyboard

8 years agomount: use the same fstab extension option syntax everywhere
Lennart Poettering [Tue, 22 May 2012 00:00:53 +0000 (02:00 +0200)]
mount: use the same fstab extension option syntax everywhere

The man page and the actual code differed, and this is now corrected

8 years agoservice: fix auto-restart handling in service_start()
David Ward [Sun, 29 Apr 2012 13:35:51 +0000 (09:35 -0400)]
service: fix auto-restart handling in service_start()

Do not bypass the SERVICE_DEAD/SERVICE_FAILED state by manually issuing
a start job while an automatic restart is pending.

8 years agoservice: fix auto-restart handling in service_stop()
David Ward [Sun, 29 Apr 2012 13:35:48 +0000 (09:35 -0400)]
service: fix auto-restart handling in service_stop()

When service_stop() handles a service in the SERVICE_AUTO_RESTART state,
it calls service_set_state() to transition it to the SERVICE_DEAD state.
However if the service failed, it should transition it to SERVICE_FAILED
instead, which will trigger its OnFailure units. To achieve this, we now
call service_enter_dead() in place of service_set_state(), which will
transition the service to either SERVICE_DEAD or SERVICE_FAILED as is

Also, some misleading comments are adjusted: service_stop() is not only
called on a user request, but also during an automatic restart in order
to handle dependencies.


8 years agoservice: actually delay auto-restart if another job is pending
David Ward [Sun, 29 Apr 2012 13:35:40 +0000 (09:35 -0400)]
service: actually delay auto-restart if another job is pending

8 years agocgtop: change default depth to 3
Lennart Poettering [Mon, 21 May 2012 23:48:40 +0000 (01:48 +0200)]
cgtop: change default depth to 3

8 years agoFixed handling of posix_fallocate() returned value
Guillermo Vidal [Wed, 9 May 2012 18:43:34 +0000 (13:43 -0500)]
Fixed handling of posix_fallocate() returned value

According to the man pages of posix_fallocate, it returns zero on
success or an error number on failure; however,  errno is not set
on failure. If the kernel or a library other than glibc does not
support the function for example, EOPNOTSUPP will be returned and
the error will not be handled properly with original code.

8 years agosystemd-analyze: switch to python getopt for argument parsing
Shawn Landden [Thu, 3 May 2012 23:41:40 +0000 (16:41 -0700)]
systemd-analyze: switch to python getopt for argument parsing

this uses gnu style getopt, so you can put the opts at the end: (e.g.)
systemd-analyze blame --user


8 years agoman: relax wording in journal-fields(7) a bit
Lennart Poettering [Mon, 21 May 2012 23:12:46 +0000 (01:12 +0200)]
man: relax wording in journal-fields(7) a bit

8 years agomain: allow system wide limits for services
Frederic Crozat [Wed, 21 Mar 2012 17:03:40 +0000 (18:03 +0100)]
main: allow system wide limits for services

8 years agojournald: one more SYSLOG_IDENTIFIER length fix
Michal Schmidt [Mon, 21 May 2012 18:21:49 +0000 (20:21 +0200)]
journald: one more SYSLOG_IDENTIFIER length fix

8 years agojournald: fix length of "SYSLOG_IDENTIFIER="
Michal Schmidt [Mon, 21 May 2012 17:31:41 +0000 (19:31 +0200)]
journald: fix length of "SYSLOG_IDENTIFIER="

Fixes weird messages like:
May 20 22:23:30 jik2 TIFIER=systemd-logind[795]: New session 46 of user gdm.

8 years agoutil: rework in_initrd() logic
Lennart Poettering [Mon, 21 May 2012 18:00:58 +0000 (20:00 +0200)]
util: rework in_initrd() logic

Checking the device major/minor is not a good idea. Let's replace this
with an explicit flag file, which we model after /etc/os-release and
call /etc/initrd-release.

8 years agomain: rename a few fix to follow general naming style
Lennart Poettering [Mon, 21 May 2012 17:48:04 +0000 (19:48 +0200)]
main: rename a few fix to follow general naming style

8 years agoswitch-root: move switch_root() call into its own .c file
Lennart Poettering [Mon, 21 May 2012 17:33:39 +0000 (19:33 +0200)]
switch-root: move switch_root() call into its own .c file

8 years agomain: do_switch_root() do not recursively remove across device boundaries
Harald Hoyer [Wed, 16 May 2012 13:08:28 +0000 (15:08 +0200)]
main: do_switch_root() do not recursively remove across device boundaries

8 years agoutil: rm_rf_children() add root_dev parameter
Harald Hoyer [Wed, 16 May 2012 13:08:27 +0000 (15:08 +0200)]
util: rm_rf_children() add root_dev parameter

if root_dev is set, remove subdirectories only, if the device is the
same as the root_dev. This prevents to remove files across device

8 years agodelta/delta.c: initialize bottom for fail state
Harald Hoyer [Wed, 16 May 2012 12:22:45 +0000 (14:22 +0200)]
delta/delta.c: initialize bottom for fail state

8 years agocore/main.c: handle the initrd timestamp differently, if in the initrd
Harald Hoyer [Wed, 16 May 2012 12:22:44 +0000 (14:22 +0200)]
core/main.c: handle the initrd timestamp differently, if in the initrd

If systemd is in the initrd/initramfs, set the initrd timestamp and do
not try to read it from the RD_TIMESTAMP environment variable.

8 years agocore/main.c: do_switch_root(): do not remove the old root if not in initrd
Harald Hoyer [Wed, 16 May 2012 12:22:43 +0000 (14:22 +0200)]
core/main.c: do_switch_root(): do not remove the old root if not in initrd

Only recursively remove the old root, if we have been in an

8 years agocore/main.c: add "--switchedroot" parameter
Harald Hoyer [Wed, 16 May 2012 12:22:42 +0000 (14:22 +0200)]
core/main.c: add "--switchedroot" parameter

If systemd serializes from a switch_root, it adds "--switchedroot" to
the systemd in the real root.

If "--switchedroot" is found, then we do not skip all the stuff, which
is skipped for normal rexecs.