chiark / gitweb /
elogind.git
5 years agolibsystemd-dhcp: revert merge into libsystemd
Tom Gundersen [Mon, 20 Jan 2014 22:40:30 +0000 (23:40 +0100)]
libsystemd-dhcp: revert merge into libsystemd

Unlike the other merged libs, the rest of libsystemd will never depend on
sd-dhcp-client, so there is no reason not to keep it separate.

5 years agoTODO: update
Kay Sievers [Mon, 20 Jan 2014 22:55:02 +0000 (23:55 +0100)]
TODO: update

5 years agoRevert "sd-bus: let sd_bus_call() use the synchronous kdbus method"
Kay Sievers [Mon, 20 Jan 2014 22:42:05 +0000 (23:42 +0100)]
Revert "sd-bus: let sd_bus_call() use the synchronous kdbus method"

This reverts commit 021b89861d0b1defcbd6ba71d1aaf6271785a942.

Something is not quite right, "KillUnit" sent from systemctl is not
handled correctly and shutdown has problems.

5 years agologind: introduce session "positions"
David Herrmann [Sat, 30 Nov 2013 10:39:48 +0000 (11:39 +0100)]
logind: introduce session "positions"

logind has no concept of session ordering. Sessions have a unique name,
some attributes about the capabilities and that's already it. There is
currently no stable+total order on sessions. If we use the logind API to
switch between sessions, we are faced with an unordered list of sessions
we have no clue of.

This used to be no problem on seats with VTs or on seats with only a
single active session. However, with the introduction of multi-session
capability for seats without VTs, we need to find a way to order sessions
in a stable way.

This patch introduces session "positions". A position is a simple integer
assigned to a session which is never changed implicitly (currently, we
also don't change it explicitly, but that may be changed someday). For
seats with VTs, we force the position to be the same as the VTnr. Without
VTs, we simply find the lowest unassigned number and use it as position.
If position-assignment fails or if, for any reason, we decide to not
assign a position to a session, the position is set to 0 (which is treated
as invalid position).
During session_load() or if two sessions have the same VTnr, we may end up
with two sessions with the same position (this shouldn't happen, but lets
be fail-safe in case some other part of the stack fails). This case is
dealt with gracefully by ignoring any session but the first session
assigned to the position. Thus, session->pos is a hint, seat->positions[i]
is the definite position-assignment. Always verify both match in case you
need to modify them!

Additionally, we introduce SwitchTo(unsigned int) on the seat-dbus-API.
You can call it with any integer value != 0 and logind will try to switch
to the request position. If you implement a compositor or any other
session-controller, you simply watch for ctrl+alt+F1 to F12 and call
SwitchTo(Fx). logind will figure a way out deal with this number.
For convenience, we also introduce SwitchToNext/Previous(). It should be
called on ctrl+alt+Left/Right (like the kernel-console used to support).

Note that the public API (SwitchTo*()) is *not* bound to the underlying
logic that is implemented now. We don't export "session-positions" on the
dbus/C API! They are an implementation detail. Instead, the SwitchTo*()
API is supposed to be a hint to let logind choose the session-switching
logic. Any foreground session-controller is free to enumerate/order
existing sessions according to their needs and call Session.Activate()
manually. But the SwitchTo*() API provides a uniform behavior across
session-controllers.

Background: Session-switching keys depend on the active keymap. The XKB
specification provides the XKB_KEY_XF86Switch_VT_1-12 key-symbols which
have to be mapped by all keymaps to allow session-switching. It is usually
bound to ctrl+alt+Fx but may be set differently. A compositor passes any
keyboard input to XKB before passing it to clients. In case a key-press
invokes the XKB_KEY_XF86Switch_VT_x action, the keypress is *not*
forwarded to clients, but instead a session-switch is scheduled.

This actually prevents us from handling these keys outside of the session.
If an active compositor has a keymap with a different mapping of these
keys, and logind itself tries to catch these combinations, we end up with
the key-press sent to the compositor's clients *and* handled by logind.
This is *bad* and we must avoid this. The only situation where a
background process is allowed to handle key-presses is debugging and
emergency-keys. In these cases, we don't care for keymap mismatches and
accept the double-event. Another exception is unmapped keys like
PowerOff/Suspend (even though this one is controversial).

5 years agoexec: introduce PrivateDevices= switch to provide services with a private /dev
Lennart Poettering [Mon, 20 Jan 2014 18:54:51 +0000 (19:54 +0100)]
exec: introduce PrivateDevices= switch to provide services with a private /dev

Similar to PrivateNetwork=, PrivateTmp= introduce PrivateDevices= that
sets up a private /dev with only the API pseudo-devices like /dev/null,
/dev/zero, /dev/random, but not any physical devices in them.

5 years agoupdate TODO
Lennart Poettering [Mon, 20 Jan 2014 16:57:42 +0000 (17:57 +0100)]
update TODO

5 years agosd-dhcp-client: refactor state machine check
Umut Tezduyar Lindskog [Mon, 20 Jan 2014 08:58:41 +0000 (09:58 +0100)]
sd-dhcp-client: refactor state machine check

5 years agosystemctl: skip native unit file handling if sysv file handling already handled every...
Lennart Poettering [Mon, 20 Jan 2014 12:43:20 +0000 (13:43 +0100)]
systemctl: skip native unit file handling if sysv file handling already handled everything

Issue pointed out by Colin Guthrie.

5 years agoupdate TODO
Lennart Poettering [Mon, 20 Jan 2014 12:42:52 +0000 (13:42 +0100)]
update TODO

5 years agosd-bus: let sd_bus_call() use the synchronous kdbus method
Daniel Mack [Sat, 18 Jan 2014 19:09:14 +0000 (20:09 +0100)]
sd-bus: let sd_bus_call() use the synchronous kdbus method

5 years agobus-kernel: move bus_kernel_make_message
Daniel Mack [Sat, 18 Jan 2014 19:06:21 +0000 (20:06 +0100)]
bus-kernel: move bus_kernel_make_message

This makes future commits more readable.

5 years agonetworkd: use 'up'/'down' rather than 'on'/'off'
Tom Gundersen [Sun, 19 Jan 2014 14:43:33 +0000 (15:43 +0100)]
networkd: use 'up'/'down' rather than 'on'/'off'

5 years agonetworkd: dhcp - by default ignore the MTU
Tom Gundersen [Sat, 18 Jan 2014 20:19:09 +0000 (21:19 +0100)]
networkd: dhcp - by default ignore the MTU

5 years agosd-dhcp-client: refactor client_{free,new}
Tom Gundersen [Sat, 18 Jan 2014 18:32:45 +0000 (19:32 +0100)]
sd-dhcp-client: refactor client_{free,new}

Make them more simiar to sd_bus and friends. Also factor out the event attachment. In the future,
we will likely want to support external main-loops, so this is a first step. For the time being,
we are still requiring an sd_event to be attached though.

5 years agosd-resolv: declare functions from .h public
Daniel Buch [Sat, 18 Jan 2014 15:07:31 +0000 (16:07 +0100)]
sd-resolv: declare functions from .h public

5 years agosd-resolve: rename get_next() and get_n_queries() + cleanup
Daniel Buch [Sat, 18 Jan 2014 13:30:28 +0000 (14:30 +0100)]
sd-resolve: rename get_next() and get_n_queries() + cleanup

5 years agosd-resolve: Allocate objects with new()
Daniel Buch [Sat, 18 Jan 2014 13:30:27 +0000 (14:30 +0100)]
sd-resolve: Allocate objects with new()

5 years agosd-resolve: get rid of sd_resolve_freeanswer()
Daniel Buch [Sat, 18 Jan 2014 13:30:26 +0000 (14:30 +0100)]
sd-resolve: get rid of sd_resolve_freeanswer()

5 years agosd-dhcp-client/networkd: add domainname support
Tom Gundersen [Sat, 18 Jan 2014 14:47:57 +0000 (15:47 +0100)]
sd-dhcp-client/networkd: add domainname support

5 years agonetworkd: don't hard depend on system bus
Tom Gundersen [Sat, 18 Jan 2014 00:37:35 +0000 (01:37 +0100)]
networkd: don't hard depend on system bus

We may not have a dbus daemon in the initrd (until we can rely on kdbus). In
this case, simply ignore any attempts at using the bus. There is only one user
for now, but surely more to come.

In order to work reliably in the real root without kdbus, but at the same time
don't delay boot when kdbus is in use, order ourselves after dbus.service.

5 years agonetworkd: resolv.conf - reword comment
Tom Gundersen [Fri, 17 Jan 2014 16:09:56 +0000 (17:09 +0100)]
networkd: resolv.conf - reword comment

Take into account that users may want to use resolvconf(8), or similar. Also,
avoid repeated calls to fputs().

5 years agolibsystemd: sync kdbus.h
Daniel Mack [Fri, 17 Jan 2014 23:50:37 +0000 (00:50 +0100)]
libsystemd: sync kdbus.h

5 years agounits: drop [Install] section from multi-user.target and graphical.target
Lennart Poettering [Fri, 17 Jan 2014 19:27:35 +0000 (20:27 +0100)]
units: drop [Install] section from multi-user.target and graphical.target

They were supposed to make it easy to make the default.target a symlink
to these targets, but this was never advertised and we have a better
command for this now in "systemctl set-default". Since the install
section makes the output of "systemctl list-unit-files" confusing (since
it makes the units appear as "disabled"), let's drop the sections.

5 years agoUpdate TODO
Lennart Poettering [Fri, 17 Jan 2014 18:59:50 +0000 (19:59 +0100)]
Update TODO

5 years agosd-resolve: rename structs to fit coding-style e.g 'struct MixedCase'
Daniel Buch [Wed, 15 Jan 2014 17:21:30 +0000 (18:21 +0100)]
sd-resolve: rename structs to fit coding-style e.g 'struct MixedCase'

lets get this right once, and if not for all, atleast for now :)

So comments and input about nameing is very welcome.

Cheers

5 years agosd-resolve: get rid of _t post-fixes for sd_ objects
Daniel Buch [Wed, 15 Jan 2014 17:21:29 +0000 (18:21 +0100)]
sd-resolve: get rid of _t post-fixes for sd_ objects

We dont want _t prefixes, right?

5 years agobuild: Skip .busname generator when kdbus is off
Umut Tezduyar Lindskog [Fri, 17 Jan 2014 12:34:35 +0000 (13:34 +0100)]
build: Skip .busname generator when kdbus is off

5 years agobus: Fix read_word_le() function
Lukasz Skalski [Fri, 17 Jan 2014 17:11:07 +0000 (18:11 +0100)]
bus: Fix read_word_le() function

5 years agoupdate TODO
Lennart Poettering [Tue, 5 Nov 2013 10:53:09 +0000 (11:53 +0100)]
update TODO

5 years agolibsystemd: rename LIBSYSTEMD_BUS to LIBSYSTEMD
Tom Gundersen [Fri, 17 Jan 2014 12:43:25 +0000 (13:43 +0100)]
libsystemd: rename LIBSYSTEMD_BUS to LIBSYSTEMD

5 years agobus: update kdbus.h (ABI break)
Kay Sievers [Fri, 17 Jan 2014 00:29:48 +0000 (01:29 +0100)]
bus: update kdbus.h (ABI break)

5 years agonetworkd: resolv.conf - extend the comment a bit
Tom Gundersen [Thu, 16 Jan 2014 19:20:08 +0000 (20:20 +0100)]
networkd: resolv.conf - extend the comment a bit

5 years agosd-dhcp-client: minor style fixes
Tom Gundersen [Thu, 16 Jan 2014 18:55:25 +0000 (19:55 +0100)]
sd-dhcp-client: minor style fixes

Also introduce a cleanup macro for DHCPLease

5 years agosd-dhcp-client: refactor DNS support
Tom Gundersen [Thu, 16 Jan 2014 18:32:22 +0000 (19:32 +0100)]
sd-dhcp-client: refactor DNS support

Rather than keeping an array of pointers to addresses, just keep an array of addresses.

5 years agosd-dhcp-client/networkd: add transient hostname support
Tom Gundersen [Mon, 13 Jan 2014 22:48:28 +0000 (23:48 +0100)]
sd-dhcp-client/networkd: add transient hostname support

5 years agonetworkd: dhcp - only set the MTU option once
Tom Gundersen [Thu, 16 Jan 2014 15:50:19 +0000 (16:50 +0100)]
networkd: dhcp - only set the MTU option once

Only set MTU request when creating the dhcp client, not every time it is restarted.

5 years agonetworkd: link - only save original MTU when necessary
Tom Gundersen [Thu, 16 Jan 2014 14:09:57 +0000 (15:09 +0100)]
networkd: link - only save original MTU when necessary

5 years agonetworkd: improve logging
Tom Gundersen [Thu, 16 Jan 2014 13:59:26 +0000 (14:59 +0100)]
networkd: improve logging

5 years agobus: fix issue with reference counting
Zbigniew Jędrzejewski-Szmek [Thu, 16 Jan 2014 06:15:17 +0000 (01:15 -0500)]
bus: fix issue with reference counting

b7fc42e03 introduced a regression.

5 years agobuild-sys: increase valgrind stack limit so all tests pass again
Zbigniew Jędrzejewski-Szmek [Tue, 14 Jan 2014 19:27:48 +0000 (14:27 -0500)]
build-sys: increase valgrind stack limit so all tests pass again

5 years agobus: break reference cycle between bus and messages
Zbigniew Jędrzejewski-Szmek [Tue, 14 Jan 2014 19:26:37 +0000 (14:26 -0500)]
bus: break reference cycle between bus and messages

Previously (6ee4f99 bus: break reference cycle between bus and
messages) I committed the test code, but not the actual fix :)

5 years agobuild-sys: fix --enable-dbus
Chengwei Yang [Wed, 15 Jan 2014 05:06:00 +0000 (13:06 +0800)]
build-sys: fix --enable-dbus

The incorrect shell code of AS_IF always get false value, so it always
build without dbus though build with "--enable-dbus" explicitely.

5 years agosd-dhcp-client/networkd: add interface MTU support
Tom Gundersen [Mon, 13 Jan 2014 22:07:59 +0000 (23:07 +0100)]
sd-dhcp-client/networkd: add interface MTU support

5 years agosd-rtnl: only alter link flags when reqeusted to
Tom Gundersen [Thu, 16 Jan 2014 00:34:27 +0000 (01:34 +0100)]
sd-rtnl: only alter link flags when reqeusted to

5 years agosd-dhcp-client: fix memory corruption
Tom Gundersen [Thu, 16 Jan 2014 00:40:23 +0000 (01:40 +0100)]
sd-dhcp-client: fix memory corruption

5 years agoevent: extend priority raneg to 64bit
Lennart Poettering [Wed, 15 Jan 2014 08:08:20 +0000 (16:08 +0800)]
event: extend priority raneg to 64bit

The kdbus prioq logic will use 64bit prios too, so let's keep this
similar in style. Using 64bit here has the advantage, that pointers and
pretty much anything else can be mapped naturally to priorities if so
desired.

5 years agosystemctl: introduce new "import-environment" command
Lennart Poettering [Wed, 15 Jan 2014 07:21:29 +0000 (15:21 +0800)]
systemctl: introduce new "import-environment" command

This may be used in graphical session start-up scripts to upload
environment variables such as $DISPLAY into the systemd manager easily.

5 years agologind: get rid of X11 display socket symlink
Lennart Poettering [Wed, 15 Jan 2014 07:19:17 +0000 (15:19 +0800)]
logind: get rid of X11 display socket symlink

X11 never made use of it anyway and it's probably better to just push
$DISPLAY into the systemd daemon from gnome-session (or equivalent
program) than to change libX11 to look for this socket. In particular
since we won't need this for Wayland anyway and we shouldn't add
infrastructure for stuff that's on its way out anyway.

5 years agoshell-completion: remove load from systemctl
Lukas Nykryn [Wed, 15 Jan 2014 14:31:07 +0000 (15:31 +0100)]
shell-completion: remove load from systemctl

5 years agoRevert "Revert "coredumpctl: in case of error free pattern after print""
Lukas Nykryn [Wed, 15 Jan 2014 11:46:44 +0000 (12:46 +0100)]
Revert "Revert "coredumpctl: in case of error free pattern after print""

In the case of the error set_consume will free the pointer 'pattern'
which is later used in log_error. Either we should stop priniting that
or use simple set_put and free pattern manually.

This reverts commit ece6b8fd5bbc1fee16f652e680e3033f2f3efc4a.

5 years agosd-resolv: rename to sd-resolve
Tom Gundersen [Tue, 14 Jan 2014 17:18:43 +0000 (18:18 +0100)]
sd-resolv: rename to sd-resolve

Lennart pointed out that we were misspelling 'resolve'. Let's not repeat the mistakes of 'umount'
and 'resolv.conf'.

5 years agoREADME: list CONFIG_FHANDLE in main section
Kay Sievers [Tue, 14 Jan 2014 14:58:04 +0000 (22:58 +0800)]
README: list CONFIG_FHANDLE in main section

5 years agoreadme: CONFIG_FHANDLE is a requirement
Umut Tezduyar Lindskog [Tue, 14 Jan 2014 14:26:54 +0000 (15:26 +0100)]
readme: CONFIG_FHANDLE is a requirement

5 years agoRevert "coredumpctl: in case of error free pattern after print"
Zbigniew Jędrzejewski-Szmek [Tue, 14 Jan 2014 14:49:10 +0000 (09:49 -0500)]
Revert "coredumpctl: in case of error free pattern after print"

This reverts commit cf50a55277ee432e1217ad894afcbd9a664ff96c.

set_consume frees on error.

5 years agocoredumpctl: in case of error free pattern after print
Lukas Nykryn [Tue, 14 Jan 2014 14:40:16 +0000 (15:40 +0100)]
coredumpctl: in case of error free pattern after print

5 years agosd-resolv: remove left-over comment
Tom Gundersen [Tue, 14 Jan 2014 12:51:18 +0000 (13:51 +0100)]
sd-resolv: remove left-over comment

We are anyway breaking the API by renaming everything, so that is not a concern.

5 years agosd-resolv: require SOCK_CLOEXEC
Tom Gundersen [Tue, 14 Jan 2014 12:50:35 +0000 (13:50 +0100)]
sd-resolv: require SOCK_CLOEXEC

5 years agosd-resolv: rename old asyncns variables
Daniel Buch [Tue, 14 Jan 2014 10:46:49 +0000 (11:46 +0100)]
sd-resolv: rename old asyncns variables

5 years agosd_resolv: Rename structs- and function names to sd_resolv
Daniel Buch [Tue, 14 Jan 2014 10:46:48 +0000 (11:46 +0100)]
sd_resolv: Rename structs- and function names to sd_resolv

5 years agobus: update kdbus.h (ABI break)
Kay Sievers [Tue, 14 Jan 2014 12:15:25 +0000 (20:15 +0800)]
bus: update kdbus.h (ABI break)

5 years agobus: update kdbus.h (ABI break)
Kay Sievers [Tue, 14 Jan 2014 08:10:08 +0000 (16:10 +0800)]
bus: update kdbus.h (ABI break)

5 years agonetworkd: DHCPv4 - allow opting out of using DNS servers
Tom Gundersen [Mon, 13 Jan 2014 20:28:53 +0000 (21:28 +0100)]
networkd: DHCPv4 - allow opting out of using DNS servers

Setting UseDNS=no will ignore any received DNS servers.

5 years agolibsystemd-dns: merge into libsystemd
Tom Gundersen [Mon, 13 Jan 2014 20:03:28 +0000 (21:03 +0100)]
libsystemd-dns: merge into libsystemd

Also rename sd-dns -> sd-resolv.

5 years agolibsystemd-rtnl: merge into libsystemd
Tom Gundersen [Mon, 13 Jan 2014 19:14:44 +0000 (20:14 +0100)]
libsystemd-rtnl: merge into libsystemd

5 years agolibsystemd-dhcp: merge into libsystemd
Tom Gundersen [Mon, 13 Jan 2014 18:12:16 +0000 (19:12 +0100)]
libsystemd-dhcp: merge into libsystemd

5 years agolibsystemd-bus: rename to libsystemd
Tom Gundersen [Mon, 13 Jan 2014 16:30:51 +0000 (17:30 +0100)]
libsystemd-bus: rename to libsystemd

Documentation was updated to refer to either 'libsystemd' or 'sd-bus' in place
of libsystemd-bus.

5 years agoTODO: update
Kay Sievers [Mon, 13 Jan 2014 02:51:26 +0000 (10:51 +0800)]
TODO: update

5 years agosd-dns: coding style cleanup
Daniel Buch [Sun, 12 Jan 2014 18:42:33 +0000 (19:42 +0100)]
sd-dns: coding style cleanup

5 years agonetworkd: fix wording
Tom Gundersen [Sun, 12 Jan 2014 14:44:40 +0000 (15:44 +0100)]
networkd: fix wording

5 years agoTODO: update rtnl/network sections
Tom Gundersen [Sat, 11 Jan 2014 20:19:10 +0000 (20:19 +0000)]
TODO: update rtnl/network sections

5 years agonetworkd: generate resolv.conf
Tom Gundersen [Sun, 5 Jan 2014 22:01:10 +0000 (23:01 +0100)]
networkd: generate resolv.conf

This adds support to generate a basic resolv.conf in /run/systemd/network.
This file will not take any effect unless a symlink is created from
/etc/resolv.conf.

Nameservers received over DHCP takes precedence over statically configured ones.

Note: /etc/resolv.conf is severely limited, so in the future we will likely
rather provide a much more powerfull nss plugin (or something to that effect),
but this should allow current users to function without any loss of
functionality.

5 years agonetworkd: bridge - remove redundant state
Tom Gundersen [Sat, 11 Jan 2014 20:20:14 +0000 (20:20 +0000)]
networkd: bridge - remove redundant state

We will not insist on getting the reply from rtnl that the bridge
was created before considering the bridge ready, as we will be
notified about that via udev. We will listen for the rtnl response
however, in case the creation of the bridge failed.

5 years agonetworkd: don't automatically reload config at runtime
Tom Gundersen [Sat, 11 Jan 2014 20:16:13 +0000 (20:16 +0000)]
networkd: don't automatically reload config at runtime

We don't know if the config will be consistent, so do as systemd itself and only
load config when the daemon starts (and possibly, in the future, when explicitly requested).

5 years agonetworkd: improve logging a bit
Tom Gundersen [Sun, 12 Jan 2014 14:24:11 +0000 (15:24 +0100)]
networkd: improve logging a bit

5 years agobus-proxyd: add some more debugging
Tom Gundersen [Fri, 10 Jan 2014 20:40:19 +0000 (20:40 +0000)]
bus-proxyd: add some more debugging

Make sure we print a message when exiting with an error.

5 years agocore: don't allow setting NOTIFY_SOCKET and similar env vars we need ourselves via...
Lennart Poettering [Sun, 12 Jan 2014 12:10:40 +0000 (20:10 +0800)]
core: don't allow setting NOTIFY_SOCKET and similar env vars we need ourselves via SetEnvironment bus calls

We just quietly eat them up, so that simple environment importing still
works without error.

5 years agoupdate TODO
Lennart Poettering [Sun, 12 Jan 2014 12:10:33 +0000 (20:10 +0800)]
update TODO

5 years agocore: clean up environment block for --user instances a bit
Lennart Poettering [Sun, 12 Jan 2014 11:39:56 +0000 (19:39 +0800)]
core: clean up environment block for --user instances a bit

5 years agobus: add test case for signature type
Lennart Poettering [Sun, 12 Jan 2014 11:15:29 +0000 (19:15 +0800)]
bus: add test case for signature type

5 years agoman: add busctl(1)
Zbigniew Jędrzejewski-Szmek [Sun, 12 Jan 2014 04:44:34 +0000 (23:44 -0500)]
man: add busctl(1)

5 years agobus: properly handle EOF error conditions in proxyd
Lennart Poettering [Sun, 12 Jan 2014 02:56:40 +0000 (10:56 +0800)]
bus: properly handle EOF error conditions in proxyd

EOF is not an error so we should not print an error message about it.

5 years agojournald: do not free space when disk space runs low
Zbigniew Jędrzejewski-Szmek [Wed, 13 Nov 2013 05:42:22 +0000 (00:42 -0500)]
journald: do not free space when disk space runs low

Before, journald would remove journal files until both MaxUse= and
KeepFree= settings would be satisfied. The first one depends (if set
automatically) on the size of the file system and is constant.  But
the second one depends on current use of the file system, and a spike
in disk usage would cause journald to delete journal files, trying to
reach usage which would leave 15% of the disk free. This behaviour is
surprising for the user who doesn't expect his logs to be purged when
disk usage goes above 85%, which on a large disk could be some
gigabytes from being full. In addition attempting to keep 15% free
provides an attack vector where filling the disk sufficiently disposes
of almost all logs.

Instead, obey KeepFree= only as a limit on adding additional files.
When replacing old files with new, ignore KeepFree=. This means that
if journal disk usage reached some high point that at some later point
start to violate the KeepFree= constraint, journald will not add files
to go above this point, but it will stay (slightly) below it. When
journald is restarted, it forgets the previous maximum usage value,
and sets the limit based on the current usage, so if disk remains to
be filled, journald might use one journal-file-size less on each
restart, if restarts happen just after rotation. This seems like a
reasonable compromise between implementation complexity and robustness.

5 years agogpt-auto-generator: use EBADSLT code when unable to detect partition type
Zbigniew Jędrzejewski-Szmek [Sat, 11 Jan 2014 21:45:29 +0000 (16:45 -0500)]
gpt-auto-generator: use EBADSLT code when unable to detect partition type

ENODEV suggests that something is missing, which is be misleading
here.

5 years agogpt-auto-generator: skip nonexistent devices
Łukasz Stelmach [Tue, 7 Jan 2014 14:00:22 +0000 (15:00 +0100)]
gpt-auto-generator: skip nonexistent devices

The devices we work with have eMMC chips for storage. The chips
provide four "hardware" partitions.  The first is /dev/mmcblk0, it
takes almost whole space and holds a GPT with several real partitions
(/dev/mmcblk0p?). Then there are three block devices (mmcblk0boot0,
mmcblk0boot1, rpmb) that are part of the same hardware as mmcblk0 that
are presented by the kernel as children of the latter. That relationship
makes gpt-auto-generator try to peek them but since they are not GPT
partitions blkid_do_safeprobe() returns -2 making verify_gpt_parition()
function return -ENODEV.

5 years agoycm: autocompose flags from the Makefile
Dave Reisner [Sun, 24 Nov 2013 13:56:56 +0000 (08:56 -0500)]
ycm: autocompose flags from the Makefile

This should hopefully be self-maintaining.

5 years agopam: skip DBUS_SESSION_BUS_ADDRESS export when kdbus is not active
Kay Sievers [Sat, 11 Jan 2014 17:18:41 +0000 (01:18 +0800)]
pam: skip DBUS_SESSION_BUS_ADDRESS export when kdbus is not active

5 years agomake socket_instantiate_service use cleanup gcc attribute
Michael Scherer [Sat, 11 Jan 2014 16:20:57 +0000 (17:20 +0100)]
make socket_instantiate_service use cleanup gcc attribute

5 years agobus: fix length calculation when serializing SD_BUS_TYPE_SIGNATURE
Kay Sievers [Sat, 11 Jan 2014 16:30:49 +0000 (00:30 +0800)]
bus: fix length calculation when serializing SD_BUS_TYPE_SIGNATURE

5 years agocore: fix unused variable
Zbigniew Jędrzejewski-Szmek [Sat, 11 Jan 2014 14:52:48 +0000 (09:52 -0500)]
core: fix unused variable

5 years agonetwork: use GNU-ism to simplify macros
Zbigniew Jędrzejewski-Szmek [Sat, 11 Jan 2014 14:42:55 +0000 (09:42 -0500)]
network: use GNU-ism to simplify macros

Thanks David!

5 years agojournalctl: allow globbing in --unit and --user-unit
Zbigniew Jędrzejewski-Szmek [Sun, 29 Dec 2013 00:47:36 +0000 (19:47 -0500)]
journalctl: allow globbing in --unit and --user-unit

This is a continuation of e3e0314b systemctl: allow globbing in commands
which take multiple unit names.

Multiple patterns can be specified, as separate arguments, or as one argument
with patterns seperated by commas.

If patterns are given, at least one unit must be matched (by any of the patterns).
This is different behaviour than systemctl, but here it is necessary because
otherwise anything would be matched, which is unlikely to be the intended
behaviour.

https://bugs.freedesktop.org/show_bug.cgi?id=59336

5 years agojournal: fix access to munmapped memory in sd_journal_enumerate_unique
Zbigniew Jędrzejewski-Szmek [Sun, 29 Dec 2013 00:33:23 +0000 (19:33 -0500)]
journal: fix access to munmapped memory in sd_journal_enumerate_unique

sd_j_e_u needs to keep a reference to an object while comparing it
with possibly duplicate objects in other files. Because the size of
mmap cache is limited, with enough files and object to compare to,
at some point the object being compared would be munmapped, resulting
in a segmentation fault.

Fix this issue by turning keep_always into a reference count that can
be increased and decreased. Other callers which set keep_always=true
are unmodified: their references are never released but are ignored
when the whole file is closed, which happens at some point. keep_always
is increased in sd_j_e_u and later on released.

5 years agosd-dns: simplify headers
Tom Gundersen [Sat, 11 Jan 2014 14:59:56 +0000 (14:59 +0000)]
sd-dns: simplify headers

As noted by Lennart: HAVE_CONFIG_H is done implicitly and sys/prctl.h is
mandatory.

5 years agosd-dns: update/add license texts
Tom Gundersen [Sat, 11 Jan 2014 14:54:22 +0000 (14:54 +0000)]
sd-dns: update/add license texts

5 years agosd-dns: cleanup public header
Tom Gundersen [Sat, 11 Jan 2014 14:51:30 +0000 (14:51 +0000)]
sd-dns: cleanup public header

Don't include internal headers
Update license text
Use _SD_BEGIN/END_DECLARATIONS

5 years agosd-dns: initial commit
Daniel Buch [Mon, 6 Jan 2014 12:41:59 +0000 (13:41 +0100)]
sd-dns: initial commit

Origin: <http://0pointer.de/lennart/projects/libasyncns/>

[tomegun: renamed some more files asyncns -> sd-dns and moved to libsystemd-bus as
requested by Lennart]

5 years agoudev/net_id: Introduce predictable network names for Linux on System z
Hendrik Brueckner [Thu, 9 Jan 2014 10:28:12 +0000 (11:28 +0100)]
udev/net_id: Introduce predictable network names for Linux on System z

Use the bus-ID to create predicatable devices names for network interfaces
on Linux on System z instances.  The bus-ID identifies a device in the s390
channel subsystem.

Network interfaces of device type Ethernet are named as:
    enccw0.0.1234    (13 characters)
up to
    enccwff.7.ffff   (14 characters)

CTC network devices of device type SLIP, use a different prefix as follows:
    slccw0.0.1234    (13 characters)

See also Red Hat Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=870859
[tomegun: typical problem of netdevs switching names between reboots.]

5 years agohwdb: Add support for Toshiba Satellite P75-A7200 keyboard
Unai Uribarri [Sat, 11 Jan 2014 14:19:41 +0000 (09:19 -0500)]
hwdb: Add support for Toshiba Satellite P75-A7200 keyboard

5 years agoRevert "bus: export bus address also when compiled without kdbus"
Kay Sievers [Fri, 10 Jan 2014 22:22:19 +0000 (06:22 +0800)]
Revert "bus: export bus address also when compiled without kdbus"

This reverts commit 9130f2128b64de19a3b7d6db7e0d371adfd296c2.

It's too early to do that. For now we should only "break" the
--enable-kdbus case, not the normal one.

5 years agoman: clarify sd_bus_e_get_errno with NULL arg
Zbigniew Jędrzejewski-Szmek [Fri, 10 Jan 2014 14:10:18 +0000 (09:10 -0500)]
man: clarify sd_bus_e_get_errno with NULL arg