chiark / gitweb /
elogind.git
5 years agoman: update the list of commands influenced by --full
Zbigniew Jędrzejewski-Szmek [Sun, 22 Dec 2013 00:07:49 +0000 (19:07 -0500)]
man: update the list of commands influenced by --full

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=732798

5 years agoman: mention --full and --lines in systemctl status
Zbigniew Jędrzejewski-Szmek [Sun, 22 Dec 2013 00:00:08 +0000 (19:00 -0500)]
man: mention --full and --lines in systemctl status

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=732798

5 years agoutil: remove union dirent_storage
Florian Weimer [Thu, 19 Dec 2013 11:44:10 +0000 (12:44 +0100)]
util: remove union dirent_storage

5 years agojournal: replace readdir_r with readdir
Florian Weimer [Thu, 19 Dec 2013 11:42:32 +0000 (12:42 +0100)]
journal: replace readdir_r with readdir

This commit also adds error handling for failures during
directory reading.

5 years agoconf-files: replace readdir_r with readdir
Florian Weimer [Thu, 19 Dec 2013 10:33:09 +0000 (11:33 +0100)]
conf-files: replace readdir_r with readdir

5 years agojournald/server: replace readdir_r with readdir
Florian Weimer [Thu, 19 Dec 2013 11:23:12 +0000 (12:23 +0100)]
journald/server: replace readdir_r with readdir

The available_space function now returns 0 if reading the directory
fails.  Previously, such errors were silently ignored.

5 years agojournal/vacuum: replace readdir_r with readdir
Florian Weimer [Thu, 19 Dec 2013 11:15:58 +0000 (12:15 +0100)]
journal/vacuum: replace readdir_r with readdir

5 years agoutil: replace readdir_r with readdir
Florian Weimer [Thu, 19 Dec 2013 11:05:41 +0000 (12:05 +0100)]
util: replace readdir_r with readdir

This fixes rm_rf_children_dangerous to detect errors during directory
reading.  Previously, it could dereference an uninitialized pointer.

5 years agoinstall: replace readdir_r with readdir
Florian Weimer [Thu, 19 Dec 2013 10:59:19 +0000 (11:59 +0100)]
install: replace readdir_r with readdir

The old code incorrectly assumed that readdir_r updates errno.

5 years agocore: replace readdir_r with readdir
Florian Weimer [Thu, 19 Dec 2013 10:25:08 +0000 (11:25 +0100)]
core: replace readdir_r with readdir

5 years agologin: replace readdir_r with readdir
Florian Weimer [Thu, 19 Dec 2013 10:14:09 +0000 (11:14 +0100)]
login: replace readdir_r with readdir

5 years agodelta: replace readdir_r with readdir
Florian Weimer [Thu, 19 Dec 2013 10:16:12 +0000 (11:16 +0100)]
delta: replace readdir_r with readdir

5 years agotmpfiles: replace readdir_r with readdir
Florian Weimer [Thu, 19 Dec 2013 11:26:07 +0000 (12:26 +0100)]
tmpfiles: replace readdir_r with readdir

5 years agologinctl: correctly show session IDs on session-status
Djalal Harouni [Sat, 21 Dec 2013 14:49:46 +0000 (15:49 +0100)]
loginctl: correctly show session IDs on session-status

Commit f8f14b3654bcd introduced a regression that makes
loginctl session-status to not show the correct session ID(s)

In print_session_status_info() the map[] array, element "Seat" receives
the offset of the "id" in "SessionStatusInfo" struct instead of the
offset of the "seat" member.

This will cause prop_map_first_of_struct() function to overwrite the
SessionStatusInfo.id memory with seats if there are any.

Fix this typo by using the "seat" member.

Before:
 - tixxdz (1000)
          Since: Sat 2013-12-21 10:07:23 CET; 5h 26min ago
         Leader: 1265 (sshd)

After:
1 - tixxdz (1000)
           Since: Sat 2013-12-21 10:07:23 CET; 5h 26min ago
          Leader: 1265 (sshd)

5 years agologinctl,shell-completions: fix listing of sessions/users/seats
Zbigniew Jędrzejewski-Szmek [Sat, 21 Dec 2013 19:20:29 +0000 (14:20 -0500)]
loginctl,shell-completions: fix listing of sessions/users/seats

5 years agobuild-sys: add xml sources to EXTRA_DIST
Zbigniew Jędrzejewski-Szmek [Sat, 21 Dec 2013 23:11:35 +0000 (18:11 -0500)]
build-sys: add xml sources to EXTRA_DIST

Apparently automake does not include the sources if they are under
a conditional that is disabled when making dist. This means that
everything would have to be enabled to make distcheck work.

5 years agoman: add systemd-dbus-proxy@.service(8) and systemd-dbus.proxy(8)
Zbigniew Jędrzejewski-Szmek [Sat, 21 Dec 2013 18:58:14 +0000 (13:58 -0500)]
man: add systemd-dbus-proxy@.service(8) and systemd-dbus.proxy(8)

5 years agobus-proxyd: use a loop instead of c&p
Zbigniew Jędrzejewski-Szmek [Sat, 21 Dec 2013 16:23:08 +0000 (11:23 -0500)]
bus-proxyd: use a loop instead of c&p

5 years agobus-proxyd: show address nicely in --help
Zbigniew Jędrzejewski-Szmek [Sat, 21 Dec 2013 15:26:55 +0000 (10:26 -0500)]
bus-proxyd: show address nicely in --help

5 years agoupdate TODO
Lennart Poettering [Sat, 21 Dec 2013 22:32:45 +0000 (23:32 +0100)]
update TODO

5 years agobus: fix bad memory access in driverd when we get an empty triggers list
Lennart Poettering [Sat, 21 Dec 2013 22:26:51 +0000 (23:26 +0100)]
bus: fix bad memory access in driverd when we get an empty triggers list

5 years agobus: when getting credentials of a bus name that is activatable but not activated...
Lennart Poettering [Sat, 21 Dec 2013 17:47:33 +0000 (18:47 +0100)]
bus: when getting credentials of a bus name that is activatable but not activated, say we don't have anything.

5 years agodriverd: properly handle NameHasOwner() for unique names
Lennart Poettering [Sat, 21 Dec 2013 17:35:27 +0000 (18:35 +0100)]
driverd: properly handle NameHasOwner() for unique names

5 years agobus: unify credential query code in driverd
Lennart Poettering [Sat, 21 Dec 2013 17:31:00 +0000 (18:31 +0100)]
bus: unify credential query code in driverd

5 years agobus: when client asks driverd for credentials of a name, return a useful error if...
Lennart Poettering [Sat, 21 Dec 2013 17:21:49 +0000 (18:21 +0100)]
bus: when client asks driverd for credentials of a name, return a useful error if that name doesn't exist on the bus

5 years agodriverd: make sure AddMatch is accessible without privileges
Lennart Poettering [Sat, 21 Dec 2013 17:13:05 +0000 (18:13 +0100)]
driverd: make sure AddMatch is accessible without privileges

5 years agoupdate TODO
Lennart Poettering [Sat, 21 Dec 2013 17:13:01 +0000 (18:13 +0100)]
update TODO

5 years agobus: when introspecting, turn unprivileged flag into inverse annoation of "privileged"
Lennart Poettering [Sat, 21 Dec 2013 17:08:39 +0000 (18:08 +0100)]
bus: when introspecting, turn unprivileged flag into inverse annoation of "privileged"

Internally, it makes sense to have a default of "privileged" for
methods, and a flag to open it up. However, externally in the bus
introspection turn this around since negative options actually suck.

5 years agobus: poll() on the right fds in proxyd
Lennart Poettering [Sat, 21 Dec 2013 16:46:35 +0000 (17:46 +0100)]
bus: poll() on the right fds in proxyd

5 years agobus: send status message from proxyd to systemd about client we are working for
Lennart Poettering [Sat, 21 Dec 2013 03:19:51 +0000 (04:19 +0100)]
bus: send status message from proxyd to systemd about client we are working for

5 years agosystemctl: fix bad memory access when processing PIDs on the "systemctl status" comma...
Lennart Poettering [Sat, 21 Dec 2013 04:07:44 +0000 (05:07 +0100)]
systemctl: fix bad memory access when processing PIDs on the "systemctl status" command line

5 years agoDEFAULT_PATH_SPLIT_USR macro
Shawn Landden [Fri, 20 Dec 2013 23:35:38 +0000 (15:35 -0800)]
DEFAULT_PATH_SPLIT_USR macro

5 years agologging: reduce send timeout to something more sensible
Zbigniew Jędrzejewski-Szmek [Sat, 14 Dec 2013 16:54:26 +0000 (11:54 -0500)]
logging: reduce send timeout to something more sensible

For a user, the timeout of 1 min per message seems equivalent to a hang.
If journald cannot process a message from PID1 for 10 ms then something
is significantly wrong. It's better to lose the message and continue.

5 years agobuild-sys: add sd_j_open_container manpage alias
Zbigniew Jędrzejewski-Szmek [Sat, 21 Dec 2013 01:26:08 +0000 (20:26 -0500)]
build-sys: add sd_j_open_container manpage alias

5 years agodriverd: sort list of names returned by ListNames
Lennart Poettering [Sat, 21 Dec 2013 03:06:11 +0000 (04:06 +0100)]
driverd: sort list of names returned by ListNames

5 years agobus: properly react to Disconnected messages in bus-proxyd
Lennart Poettering [Sat, 21 Dec 2013 03:03:12 +0000 (04:03 +0100)]
bus: properly react to Disconnected messages in bus-proxyd

5 years agobus: change bus-proxyd command line parsing to be more similar to other tools
Lennart Poettering [Sat, 21 Dec 2013 02:59:07 +0000 (03:59 +0100)]
bus: change bus-proxyd command line parsing to be more similar to other tools

5 years agobus: refuse messages pretending to originate from local interface
Lennart Poettering [Sat, 21 Dec 2013 02:52:39 +0000 (03:52 +0100)]
bus: refuse messages pretending to originate from local interface

5 years agobus: in sd_bus_try_close() consider local read queue too
Lennart Poettering [Sat, 21 Dec 2013 02:46:26 +0000 (03:46 +0100)]
bus: in sd_bus_try_close() consider local read queue too

5 years agobus: always override message destination field with data from kdbus
Lennart Poettering [Sat, 21 Dec 2013 02:17:29 +0000 (03:17 +0100)]
bus: always override message destination field with data from kdbus

5 years agobus: always consider well-known names sender credentials as attached to messages
Lennart Poettering [Sat, 21 Dec 2013 02:16:39 +0000 (03:16 +0100)]
bus: always consider well-known names sender credentials as attached to messages

5 years agobus: when filtering by sender take well known names lest from attached creds into...
Lennart Poettering [Sat, 21 Dec 2013 01:56:34 +0000 (02:56 +0100)]
bus: when filtering by sender take well known names lest from attached creds into account

5 years agocore: make check for Disconnected message more precise
Lennart Poettering [Sat, 21 Dec 2013 01:40:40 +0000 (02:40 +0100)]
core: make check for Disconnected message more precise

5 years agobus: when watching bus names, we always need to add a change match
Lennart Poettering [Sat, 21 Dec 2013 00:02:23 +0000 (01:02 +0100)]
bus: when watching bus names, we always need to add a change match

Since for the kernel change of ownership of a name between starter and
real client is a change and for userspace thats addition/removal, we
need to unconditionally subscribe to these change events.

5 years agocore: pass notify fd across reexecs
Lennart Poettering [Fri, 20 Dec 2013 23:19:37 +0000 (00:19 +0100)]
core: pass notify fd across reexecs

That way we the random socket name stays stable across reexec and we
won't lose client messages.

5 years agoupdate TODO
Lennart Poettering [Fri, 20 Dec 2013 23:19:30 +0000 (00:19 +0100)]
update TODO

5 years agobus: wrap KDBUS_CMD_BYEBYE with a high-level call sd_bus_try_close()
Lennart Poettering [Fri, 20 Dec 2013 22:25:43 +0000 (23:25 +0100)]
bus: wrap KDBUS_CMD_BYEBYE with a high-level call sd_bus_try_close()

5 years agolibsystemd-bus: match on any connection ID unless specified
Daniel Mack [Fri, 20 Dec 2013 22:12:42 +0000 (23:12 +0100)]
libsystemd-bus: match on any connection ID unless specified

5 years agolibsystemd-bus: follow changes in kdbus
Daniel Mack [Fri, 20 Dec 2013 19:57:20 +0000 (20:57 +0100)]
libsystemd-bus: follow changes in kdbus

In particular, the match interface has seen a major overhaul.

5 years agodef: lower exit-on-idle time
Lennart Poettering [Thu, 19 Dec 2013 21:12:53 +0000 (22:12 +0100)]
def: lower exit-on-idle time

5 years agosd-event: return 1 in sd_event_run() when we hit EINTR
Lennart Poettering [Thu, 19 Dec 2013 20:38:52 +0000 (21:38 +0100)]
sd-event: return 1 in sd_event_run() when we hit EINTR

That way we can distuingish timeouts from EINTR

5 years agoservice: don't send out PropertiesChanged message for each watchdog notification
Lennart Poettering [Thu, 19 Dec 2013 20:34:09 +0000 (21:34 +0100)]
service: don't send out PropertiesChanged message for each watchdog notification

5 years agomachined: implement exit-on-idle
Lennart Poettering [Thu, 19 Dec 2013 20:15:08 +0000 (21:15 +0100)]
machined: implement exit-on-idle

5 years agobus: fix exit-on-idle for driverd
Lennart Poettering [Thu, 19 Dec 2013 20:14:52 +0000 (21:14 +0100)]
bus: fix exit-on-idle for driverd

5 years agobus: reenable id change subscriptions
Lennart Poettering [Thu, 19 Dec 2013 20:13:01 +0000 (21:13 +0100)]
bus: reenable id change subscriptions

5 years agoremove unused variable
Thomas Hindoe Paaboel Andersen [Thu, 19 Dec 2013 21:06:47 +0000 (22:06 +0100)]
remove unused variable

5 years agortnl: correctly get the size of data to be appended
Tom Gundersen [Thu, 19 Dec 2013 20:28:50 +0000 (21:28 +0100)]
rtnl: correctly get the size of data to be appended

This fixes:
  systemd[1]: Failed to configure loopback device: Numerical result out of range

5 years agoloopback-setup: correctly set flags and scope for ipv6 address
Tom Gundersen [Thu, 19 Dec 2013 19:11:34 +0000 (20:11 +0100)]
loopback-setup: correctly set flags and scope for ipv6 address

5 years agokeymap: Refactor Acer tables
Martin Pitt [Thu, 19 Dec 2013 07:33:32 +0000 (08:33 +0100)]
keymap: Refactor Acer tables

Move common keys into generic Acer table, and factorize the model specific
exceptions.

5 years agobus: do not add empty strings, size should be 0
Kay Sievers [Thu, 19 Dec 2013 05:55:44 +0000 (06:55 +0100)]
bus: do not add empty strings, size should be 0

5 years agosystemct: add empty line between units in cat
Zbigniew Jędrzejewski-Szmek [Thu, 19 Dec 2013 03:10:17 +0000 (22:10 -0500)]
systemct: add empty line between units in cat

5 years agologinctl: improve print_{session|user|seat}_status_info() functions
Djalal Harouni [Tue, 17 Dec 2013 18:42:03 +0000 (19:42 +0100)]
loginctl: improve print_{session|user|seat}_status_info() functions

1) Instead of checking if we need to print a new line on each iteration,
pass the "new_line" as a pointer to those functions, so they can use
it to check if a new line is needed. This makes the code more consistent
as it is done in other places: machinectl, systemctl...

2) Move the error messages from show_{session|user|seat}() to their
appropriate print_{session|user|seat}_status_info() functions, this will
prevent from logging an error message twice in case show_properties()
fails and it will improve code readability.

3) Also do not ignore error codes on these functions.

5 years agologinctl: use show_properties() to get login1 properties
Djalal Harouni [Tue, 17 Dec 2013 18:42:01 +0000 (19:42 +0100)]
loginctl: use show_properties() to get login1 properties

Commit f8f14b3654bcd introduced a regression that makes loginctl ignore
the "--property" option.

This patch fixes the bug, it uses a new show_properties() function to
query and filter properties.

5 years agologinctl: replace strv_append() by strv_extend()
Djalal Harouni [Tue, 17 Dec 2013 18:42:00 +0000 (19:42 +0100)]
loginctl: replace strv_append() by strv_extend()

5 years agojournalctl,zsh-completion: fix several issues in --help message text
Jason St. John [Wed, 18 Dec 2013 02:36:27 +0000 (21:36 -0500)]
journalctl,zsh-completion: fix several issues in --help message text

-- fix grammar and reword some descriptions for clarity
-- add a useful description of what --follow does
-- fix the description for --after-cursor
-- properly introduce the FSS acronym for "Forward Secure Sealing" in
both sections
-- clarify the --disk-usage command

[zj: perform similar changes to zsh completions]

squash! journalctl: fix several issues in --help message text

5 years agoman: add DOI for refereed article on Forward Secure Sealing to journald.conf(5)
Jason St. John [Wed, 18 Dec 2013 00:40:02 +0000 (19:40 -0500)]
man: add DOI for refereed article on Forward Secure Sealing to journald.conf(5)

In journalctl(1), be more explicit about the reference to "Seal=" in
journald.conf(5) and what information can be found there.

5 years agoman: be more explicit about option arguments that take 128-bit IDs in journalctl(1)
Jason St. John [Tue, 17 Dec 2013 23:48:43 +0000 (18:48 -0500)]
man: be more explicit about option arguments that take 128-bit IDs in journalctl(1)

It may not be immediately obvious to the reader what "ID128" is, so replace the
example option argument "ID128" with "128-bit-ID".

5 years agoman: fix grammar issues in journalctl(1)
Jason St. John [Tue, 17 Dec 2013 23:48:42 +0000 (18:48 -0500)]
man: fix grammar issues in journalctl(1)

And add a missing <option> tag around "--setup-keys" under "--force".

5 years agomachinectl: show_properties() already logs the error
Djalal Harouni [Tue, 17 Dec 2013 22:40:15 +0000 (23:40 +0100)]
machinectl: show_properties() already logs the error

5 years agoudevadm: modernization
Zbigniew Jędrzejewski-Szmek [Wed, 18 Dec 2013 02:49:07 +0000 (21:49 -0500)]
udevadm: modernization

5 years agoudevadm,scsi_id: add short options to help strings and to the man page
Zbigniew Jędrzejewski-Szmek [Wed, 18 Dec 2013 02:48:14 +0000 (21:48 -0500)]
udevadm,scsi_id: add short options to help strings and to the man page

Also clean things up a bit here and there.

5 years agoscsi_id: cleanup
Zbigniew Jędrzejewski-Szmek [Sun, 15 Dec 2013 22:15:54 +0000 (17:15 -0500)]
scsi_id: cleanup

Remove -i option which would case exit(1) to happen.
Remove some unused code.
Convert to bool where appropriate.
Simplify things a bit.
Always free everything.

https://bugzilla.redhat.com/show_bug.cgi?id=1043304

5 years agodriverd: implement AddMatch/RemoveMatch logic
Lennart Poettering [Thu, 19 Dec 2013 02:02:45 +0000 (03:02 +0100)]
driverd: implement AddMatch/RemoveMatch logic

5 years agodo not run kmod-setup in a container
Kay Sievers [Thu, 19 Dec 2013 00:32:55 +0000 (01:32 +0100)]
do not run kmod-setup in a container

5 years agotemporarily support "kdbus" keyword on the kernel commandline to load the module
Kay Sievers [Wed, 18 Dec 2013 22:56:35 +0000 (23:56 +0100)]
temporarily support "kdbus" keyword on the kernel commandline to load the module

5 years agolibudev: monitor - only disallow "udev" type events
Kay Sievers [Wed, 18 Dec 2013 21:30:26 +0000 (22:30 +0100)]
libudev: monitor - only disallow "udev" type events

5 years agolibudev: disable monitor inside a container
Kay Sievers [Wed, 18 Dec 2013 21:01:31 +0000 (22:01 +0100)]
libudev: disable monitor inside a container

Uevents are events of the host, which should not leak into a container.
Containers do not support hotplug at the moment, and devices and uevents
are not namespace aware.

5 years agobus: catch up with kdbus changes
Kay Sievers [Wed, 18 Dec 2013 20:16:35 +0000 (21:16 +0100)]
bus: catch up with kdbus changes

5 years agobus-driverd: fix flags translation in driver_request_name
Lukasz Skalski [Wed, 18 Dec 2013 12:32:13 +0000 (13:32 +0100)]
bus-driverd: fix flags translation in driver_request_name

Flags for a name entry (KDBUS_NAME_*) should be set for
cmd_name->flags (not conn_flags)

5 years agoRevert "ptyfwd: make master terminal attributes raw, too"
Lennart Poettering [Wed, 18 Dec 2013 18:24:00 +0000 (19:24 +0100)]
Revert "ptyfwd: make master terminal attributes raw, too"

This reverts commit e7d43b3cc30764138c90eaaf95d3d8f49e448890.

This broke the console terminal when booting up a container, so let's
not do this.

5 years agobus: reduce calls to KDBUS_CMD_MEMFD_SIZE_SET ioctl
Lennart Poettering [Wed, 18 Dec 2013 17:46:23 +0000 (18:46 +0100)]
bus: reduce calls to KDBUS_CMD_MEMFD_SIZE_SET ioctl

Instead of calling it for each buffer append, increase allocation
exponentially and set the real value only at the end, when sealing off
the memfd.

This should drastically reduce the number of times we invoke the
ioctl().

5 years agoptyfwd: make master terminal attributes raw, too
Lennart Poettering [Wed, 18 Dec 2013 17:14:10 +0000 (18:14 +0100)]
ptyfwd: make master terminal attributes raw, too

5 years agounits: when spawning a getty configure TERM explicitly
Lennart Poettering [Wed, 18 Dec 2013 16:48:31 +0000 (17:48 +0100)]
units: when spawning a getty configure TERM explicitly

This way we can make use of our logic to automatically determine an
appropriate TERM for a specific tty.

5 years agoexecute: set TERM even if we don't open the tty on our own
Lennart Poettering [Wed, 18 Dec 2013 16:41:16 +0000 (17:41 +0100)]
execute: set TERM even if we don't open the tty on our own

This way, when a tty path is configured TERM is set, which is nice to
set a useful term for gettys.

5 years agologinctl: when showing device tree of seats with no devices show something useful
Lennart Poettering [Wed, 18 Dec 2013 16:16:33 +0000 (17:16 +0100)]
loginctl: when showing device tree of seats with no devices show something useful

5 years agocore,logind: libudev usage modernizations
Lennart Poettering [Wed, 18 Dec 2013 16:13:42 +0000 (17:13 +0100)]
core,logind: libudev usage modernizations

Always use cleanup logic and don't eat up errors returned by libudev

5 years agocore,logind,networkd: check for udev device initialization via enumeration matches
Lennart Poettering [Wed, 18 Dec 2013 16:12:15 +0000 (17:12 +0100)]
core,logind,networkd: check for udev device initialization via enumeration matches

Instead of checking each device after we got it, check wuth an
enumeration filter instead, to make it more efficient.

5 years agocore: priorize notification fd processing over notification fd process via sd-event...
Lennart Poettering [Wed, 18 Dec 2013 02:46:53 +0000 (03:46 +0100)]
core: priorize notification fd processing over notification fd process via sd-event's logic

5 years agolog: when we log to /dev/console and got disconnected (maybe due to vhangup) reconnect
Lennart Poettering [Wed, 18 Dec 2013 15:49:15 +0000 (16:49 +0100)]
log: when we log to /dev/console and got disconnected (maybe due to vhangup) reconnect

5 years agounits: run systemd-networkd.service only if CAP_NET_ADMIN capability is around
Lennart Poettering [Wed, 18 Dec 2013 15:45:20 +0000 (16:45 +0100)]
units: run systemd-networkd.service only if CAP_NET_ADMIN capability is around

This has the effect that systemd-networkd won't run in containers
without network namespacing wher CAP_NET_ADMIN is (usually) not
available. It will still run in containers with network namespacing on
(where CAP_NET_ADMIN is usually avilable).

We might remove this condition check again if networkd provides services
to apps that also are useful in containers lacking network namespacing,
however, as long as it doesn't it should be handled like udevd and be
excluded in such containers.

5 years agocore: in a container log to /dev/console if "debug" is specified
Lennart Poettering [Wed, 18 Dec 2013 04:09:53 +0000 (05:09 +0100)]
core: in a container log to /dev/console if "debug" is specified

5 years agolog: don't reopen /dev/console each time we call log_open()
Lennart Poettering [Wed, 18 Dec 2013 04:07:34 +0000 (05:07 +0100)]
log: don't reopen /dev/console each time we call log_open()

Instead, force reopen it only if we really really have to.

5 years agoservice: watch main pid even in final states
Lennart Poettering [Wed, 18 Dec 2013 03:19:20 +0000 (04:19 +0100)]
service: watch main pid even in final states

In some circumstances, for example when start-up times out we
immediately jump into the final state, at which point we still should
try to watch the main pid so that the SIGCHLD allows us to quickly
move into dead state.

5 years agolibudev: devices received from udev are always initialized
Kay Sievers [Wed, 18 Dec 2013 15:49:19 +0000 (16:49 +0100)]
libudev: devices received from udev are always initialized

5 years agologin: Don't stop a running user manager from garbage-collecting the user.
Thomas Bächler [Sun, 15 Dec 2013 11:06:37 +0000 (12:06 +0100)]
login: Don't stop a running user manager from garbage-collecting the user.

With the current logic, a user will never be garbage-collected, since its
manager will always be around. Change the logic such that a user is
garbage-collected when it has no sessions and linger is disabled.

5 years agocomment out udev's is_initialized call until the problem is sorted out
Kay Sievers [Wed, 18 Dec 2013 04:06:32 +0000 (05:06 +0100)]
comment out udev's is_initialized call until the problem is sorted out

5 years agocore,logind,networkd: don't pick up devices from udev before they finished udev initi...
Lennart Poettering [Wed, 18 Dec 2013 02:37:26 +0000 (03:37 +0100)]
core,logind,networkd: don't pick up devices from udev before they finished udev initialization

Managers shouldn't pick up the devices the manage before udev finished
initialization, hence check explicitly for that.

5 years agobusctl: output a single sorted list of names, including activatable and activated
Lennart Poettering [Wed, 18 Dec 2013 01:49:03 +0000 (02:49 +0100)]
busctl: output a single sorted list of names, including activatable and activated

5 years agobus: beef up driverd
Lennart Poettering [Wed, 18 Dec 2013 01:25:19 +0000 (02:25 +0100)]
bus: beef up driverd

Add validity checks for all parameters passed in.

Implement ReleaseName and StartServiceByName().

Fix access control in vtable.

5 years agobus: when releasing, taking, listing bus names check if they are valid
Lennart Poettering [Wed, 18 Dec 2013 01:23:07 +0000 (02:23 +0100)]
bus: when releasing, taking, listing bus names check if they are valid

Given that the kernel does not validate bus names we have to when using
them.