platform/upstream/systemd.git
9 years agobus: change creds dumping order to be more close to internal storage order
Lennart Poettering [Mon, 24 Nov 2014 20:37:26 +0000 (21:37 +0100)]
bus: change creds dumping order to be more close to internal storage order

9 years agoutil: skip incomplete ucred information in getpeersec()
Lennart Poettering [Mon, 24 Nov 2014 20:21:24 +0000 (21:21 +0100)]
util: skip incomplete ucred information in getpeersec()

9 years agosd-bus: don't blindly take incomplete ucred bits from AF_UNIX when constructing message
Lennart Poettering [Mon, 24 Nov 2014 20:20:53 +0000 (21:20 +0100)]
sd-bus: don't blindly take incomplete ucred bits from AF_UNIX when constructing message

9 years agoupdate TODO
Lennart Poettering [Mon, 24 Nov 2014 16:55:44 +0000 (17:55 +0100)]
update TODO

9 years agohwdb: fix a typo
Peter Hutterer [Tue, 25 Nov 2014 10:45:50 +0000 (20:45 +1000)]
hwdb: fix a typo

9 years agotmpfiles.d: Fix directory name
Martin Pitt [Tue, 25 Nov 2014 10:39:30 +0000 (11:39 +0100)]
tmpfiles.d: Fix directory name

The .service uses "/var/lib/container", not "containers".

9 years agolocaled: verify xkb keymaps after pk-queries
David Herrmann [Tue, 25 Nov 2014 09:26:08 +0000 (10:26 +0100)]
localed: verify xkb keymaps after pk-queries

Make sure not to run xkb-keymap validation twice if pk-queries are used.
Move it below pk-checks.

9 years agoterminal/idev: forward xkb-messages
David Herrmann [Tue, 25 Nov 2014 09:24:39 +0000 (10:24 +0100)]
terminal/idev: forward xkb-messages

Properly forward all XKB messages. You can use XKB_LOG_VERBOSITY= to
control the amount of messages sent by XKB. We explicitly set
XKB_LOG_LEVEL to 7 you can use SYSTEMD_LOG_LEVEL to control the log-level
generically.

9 years agoterminal/idev: avoid magic numbers
David Herrmann [Tue, 25 Nov 2014 09:24:08 +0000 (10:24 +0100)]
terminal/idev: avoid magic numbers

Use XKB_CONTEXT_NO_FLAGS instead of magic 0.

9 years agoterminal/idev: use compose tables
David Herrmann [Mon, 24 Nov 2014 17:25:30 +0000 (18:25 +0100)]
terminal/idev: use compose tables

Before forwarding keyboard events, feed them into possible compose tables.
This enables Compose-key and Dead-key features.

Few notes:
 * REPEAT events are never fed into compose tables. It just doesn't make
   sense and is usually not wanted. Compose-sequences are usually hard to
   remember and take time to type. Thus, the REPEAT event of the
   Compose-key itself would often cancel the compose sequence already.

 * Stop resolving symbols for UP events. Anything but keycodes is never
   associated to a physical key, but is a one-time action. There is
   nothing like UP events for key-symbols!

 * Cancel compose-sequences on Multi-Key UP. See the inline comment. We
   should make this configurable!

9 years agoterminal/idev: add compose-file support
David Herrmann [Mon, 24 Nov 2014 15:32:54 +0000 (16:32 +0100)]
terminal/idev: add compose-file support

Add support for compose files to idev-keyboard. This requires
libxkbcommon-0.5.0, which is pretty new, but should be fine.

We don't use the compose-files, yet. Further commits will put life into
them.

9 years agosmack: introduce new SmackProcessLabel option
WaLyong Cho [Mon, 24 Nov 2014 11:46:20 +0000 (20:46 +0900)]
smack: introduce new SmackProcessLabel option

In service file, if the file has some of special SMACK label in
ExecStart= and systemd has no permission for the special SMACK label
then permission error will occurred. To resolve this, systemd should
be able to set its SMACK label to something accessible of ExecStart=.
So introduce new SmackProcessLabel. If label is specified with
SmackProcessLabel= then the child systemd will set its label to
that. To successfully execute the ExecStart=, accessible label should
be specified with SmackProcessLabel=.
Additionally, by SMACK policy, if the file in ExecStart= has no
SMACK64EXEC then the executed process will have given label by
SmackProcessLabel=. But if the file has SMACK64EXEC then the
SMACK64EXEC label will be overridden.

[zj: reword man page]

9 years agogpt-auto-generator: properly ignore value
Zbigniew Jędrzejewski-Szmek [Mon, 24 Nov 2014 01:01:40 +0000 (20:01 -0500)]
gpt-auto-generator: properly ignore value

A negative return code was treated as a true value.

9 years agobuild-sys: support local ./configure arguments
David Herrmann [Mon, 24 Nov 2014 14:39:00 +0000 (15:39 +0100)]
build-sys: support local ./configure arguments

I often want to use the awesome "./autogen.sh [cmd]" arguments, but have
to append some custom ./configure options. For now, I always had to edit
autogen.sh manually, or copy the full commands out of it and run it
myself.

As I think this is super annoying, this commit adds support for
".config.args" files in $topdir. If it exists, any content is just
appended to $args, thus to any ./configure invokation of autogen.sh.

Maybe autotools provide something similar out-of-the-box. In that case,
feel free to revert this and lemme know!

9 years agolocaled: validate xkb keymaps
David Herrmann [Mon, 24 Nov 2014 14:12:42 +0000 (15:12 +0100)]
localed: validate xkb keymaps

Introduce a new optional dependency on libxkbcommon for systemd-localed.
Whenever the x11 keymap settings are changed, use libxkbcommon to compile
the keymap. If the compilation fails, print a warning so users will get
notified.

On compilation failure, we still update the keymap settings for now. This
patch just introduces the xkbcommon infrastructure to have keymap
validation in place. We can later decide if/how we want to enforce this.

9 years agocryptsetup: default to no hash when keyfile is specified
Zbigniew Jędrzejewski-Szmek [Mon, 24 Nov 2014 14:11:12 +0000 (09:11 -0500)]
cryptsetup: default to no hash when keyfile is specified

For plain dm-crypt devices, the behavior of cryptsetup package is to
ignore the hash algorithm when a key file is provided. It seems wrong
to ignore a hash when it is explicitly specified, but we should default
to no hash if the keyfile is specified.

https://bugs.freedesktop.org/show_bug.cgi?id=52630

9 years agomanager: only issue overmount warning when the check succeeded
Zbigniew Jędrzejewski-Szmek [Thu, 6 Nov 2014 03:39:13 +0000 (22:39 -0500)]
manager: only issue overmount warning when the check succeeded

If for any reason the check failed (selinux?), we would still issue
the warning. Check the return status.

9 years agomanager: let manager_free() handle NULLs
Zbigniew Jędrzejewski-Szmek [Sat, 8 Nov 2014 15:06:12 +0000 (10:06 -0500)]
manager: let manager_free() handle NULLs

This makes the calling code a bit simpler.

9 years agodbus: return non-zero return value in the case that prefix won't match
Lukas Nykryn [Thu, 20 Nov 2014 15:06:18 +0000 (16:06 +0100)]
dbus: return non-zero return value in the case that prefix won't match

strv_extend returns 0 in the case of success which means that
else if (bus_track_deserialize_item(&m->deserialized_subscribed, l) == 0)
        log_warning("Unknown serialization item '%s'", l);
will be printed when value is added correctly.

9 years agotest: support empty environment variables in unit files
Iago López Galeiras [Thu, 20 Nov 2014 20:18:23 +0000 (21:18 +0100)]
test: support empty environment variables in unit files

Also update TODO, empty environment variables in Environment= and
EnvironmentFile= options work.

9 years agoman: reword list-dependencies description
Zbigniew Jędrzejewski-Szmek [Sun, 23 Nov 2014 16:07:15 +0000 (11:07 -0500)]
man: reword list-dependencies description

The previous version was a bit too vague. It is better
to simply list all dependency types that are followed.

Previous version also made an emphasis on dependencies introduced by
configuration.  But this command (or systemd) don't care about this
distinction between configured and automatically added dependencies at
all. This distinctionis removed from the main description, and an
explanatory paragraph is added to remind the user that all
dependencies are shown, no matter where they come from.

9 years agosystemctl: show BindsTo/BoundBy in list-dependencies
Zbigniew Jędrzejewski-Szmek [Sun, 23 Nov 2014 01:21:56 +0000 (20:21 -0500)]
systemctl: show BindsTo/BoundBy in list-dependencies

Suggested-by: Peter Mattern <matternp@arcor.de>
http://lists.freedesktop.org/archives/systemd-devel/2014-November/025437.html

9 years agonspawn: ignore EEXIST when mounting tmpfs
Richard Schütz [Fri, 21 Nov 2014 20:44:48 +0000 (21:44 +0100)]
nspawn: ignore EEXIST when mounting tmpfs

commit 79d80fc1466512d0ca211f4bfcd9de5f2f816a5a introduced a regression that
prevents mounting a tmpfs if the mount point already exits in the container's
root file system. This commit fixes the problem by ignoring EEXIST.

9 years agoman/logind.conf.xml: make consistent with default logind.conf
Chris Atkinson [Sat, 22 Nov 2014 20:58:47 +0000 (15:58 -0500)]
man/logind.conf.xml: make consistent with default logind.conf

https://bugs.freedesktop.org/show_bug.cgi?id=86263

9 years agoDo not check for existence of remote binaries
Zbigniew Jędrzejewski-Szmek [Sat, 22 Nov 2014 23:00:07 +0000 (18:00 -0500)]
Do not check for existence of remote binaries

systemd-run would fail when run with -M or -H and an absolute path,
if this path did not exists locally. Allow it to continue, since we
don't have a nice way of checking if the binary exists remotely.

The case where -M or -H is used and a local path is unchanged, and we
still iterate over $PATH to find the binary. We need to convert to an
absolute path, and we don't have a nice mechanism to check remotely,
so we assume that the binary will be located in the same place locally
and remotely.

http://lists.freedesktop.org/archives/systemd-devel/2014-November/025418.html

9 years agomissing: add IFLA_BRPORT_LEARNING and friends
Zbigniew Jędrzejewski-Szmek [Sat, 22 Nov 2014 22:12:44 +0000 (17:12 -0500)]
missing: add IFLA_BRPORT_LEARNING and friends

IFLA_BRPORT_LEARNING was added in v3.10-rc4-583-g9ba18891f7,
and IFLA_BRPORT_UNICAST_FLOOD in v3.10-rc4-584-g867a59436f.

9 years agomissing: add __NR_getrandom for a few more archs
Zbigniew Jędrzejewski-Szmek [Sat, 22 Nov 2014 17:04:50 +0000 (12:04 -0500)]
missing: add __NR_getrandom for a few more archs

Mips has getrandom() too, but there's just too many variants
for me too care. Either someone who cares does it, or they get
compile-time warnings with old kernel headers.

10 years agoselinux-access: fix broken ternary operator
Lukas Nykryn [Fri, 21 Nov 2014 16:02:15 +0000 (17:02 +0100)]
selinux-access: fix broken ternary operator

it seems to be a typo introduced by ebcf1f97de4f6b1580ae55eb56b1a3939fe6b602
-                _r = selinux_access_check(_b, _m, _u->source_path ?:_u->fragment_path, (permission), &_error); \
+        ({ Unit *_unit = (unit); selinux_generic_access_check(bus,message, _unit->fragment_path ?: _unit->fragment_path, permission,error); })

10 years agoman: don't mix 2ch and 8ch indentation in busctl.xml
Lennart Poettering [Fri, 21 Nov 2014 19:46:01 +0000 (20:46 +0100)]
man: don't mix 2ch and 8ch indentation in busctl.xml

10 years agoman: add emacs header to get correct indention in nxml-mode for the manpage XML files...
Lennart Poettering [Fri, 21 Nov 2014 19:44:48 +0000 (20:44 +0100)]
man: add emacs header to get correct indention in nxml-mode for the manpage XML files that use 2ch indenting

In the long run we really should figure out if we want to stick with 8ch
or 2ch indenting, and not continue with half-and-half. For now, just
make emacs aware of the files that use 2ch indenting.

10 years agobusctl: add --timeout= option to specify method call timeout
Lennart Poettering [Fri, 21 Nov 2014 19:31:50 +0000 (20:31 +0100)]
busctl: add --timeout= option to specify method call timeout

10 years agobusctl: add options to control message header flags when invoking methods
Lennart Poettering [Fri, 21 Nov 2014 19:13:26 +0000 (20:13 +0100)]
busctl: add options to control message header flags when invoking methods

10 years agonspawn: Add try-{host,guest} journal link modes
Martin Pitt [Thu, 20 Nov 2014 13:30:52 +0000 (14:30 +0100)]
nspawn: Add try-{host,guest} journal link modes

--link-journal={host,guest} fail if the host does not have persistent
journalling enabled and /var/log/journal/ does not exist. Even worse, as there
is no stdout/err any more, there is no error message to point that out.

Introduce two new modes "try-host" and "try-guest" which don't fail in this
case, and instead just silently skip the guest journal setup.

Change -j to mean "try-guest" instead of "guest", and fix the wrong --help
output for it (it said "host" before).

Change systemd-nspawn@.service.in to use "try-guest" so that this unit works
with both persistent and non-persistent journals on the host without failing.

https://bugs.debian.org/770275

10 years agounits: make sure rfkill service is bount to the actual hardware
Lennart Poettering [Fri, 21 Nov 2014 00:20:57 +0000 (01:20 +0100)]
units: make sure rfkill service is bount to the actual hardware

10 years agoRevert "systemd-logind.service: set Type=notify"
Lennart Poettering [Fri, 21 Nov 2014 00:17:52 +0000 (01:17 +0100)]
Revert "systemd-logind.service: set Type=notify"

This reverts commit a4962513c555fe3ac4b5bebf97a71701361a45b0.

logind.service is a D-Bus service, hence we should use the dbus name as
indication that we are up. Type=dbus is implied if BusName= is
specified, as it is in this case.

This removes a warning that is printed because a BusName= is specified
for a Type=notify unit.

10 years agotmpfiles.d: Create /var/lib/containers
Martin Pitt [Thu, 20 Nov 2014 13:37:08 +0000 (14:37 +0100)]
tmpfiles.d: Create /var/lib/containers

Create /var/lib/containers so that it exists with an appropriate mode. We want
0700 by default so that users on the host aren't able to call suid root
binaries in the container. This becomes a security issue if a user can enter a
container as root, create a suid root binary, and call that from the host.
(This assumes that containers are caged by mandatory access control or are
started as user).

10 years agobusctl: show property values in "introspect" output, add "set-property" command,...
Lennart Poettering [Thu, 20 Nov 2014 22:12:29 +0000 (23:12 +0100)]
busctl: show property values in "introspect" output, add "set-property" command, and support both a terse and a verbose output format

10 years agobusctl: when --address is specified do not assume we connect to a full bus
Lennart Poettering [Thu, 20 Nov 2014 20:01:06 +0000 (21:01 +0100)]
busctl: when --address is specified do not assume we connect to a full bus

10 years agobusctl: use canned error message on parse failure
Lennart Poettering [Thu, 20 Nov 2014 19:59:57 +0000 (20:59 +0100)]
busctl: use canned error message on parse failure

10 years agosd-bus: create clean error when a property Set() call with incorrect signature is...
Lennart Poettering [Thu, 20 Nov 2014 19:58:39 +0000 (20:58 +0100)]
sd-bus: create clean error when a property Set() call with incorrect signature is passed in

10 years agobuild: avoid trailing whitespace in feature string
Lennart Poettering [Thu, 20 Nov 2014 19:58:02 +0000 (20:58 +0100)]
build: avoid trailing whitespace in feature string

10 years agoutil: add functions getting proc cwd and root
Jakub Filak [Wed, 19 Nov 2014 10:01:19 +0000 (11:01 +0100)]
util: add functions getting proc cwd and root

/proc/[pid]/cwd and /proc/[pid]/root are symliks to corresponding
directories

The added functions returns values of that symlinks.

10 years agosd-dhcp-lease: fix copy-paste error asserting wrong function argument
Thomas Haller [Wed, 19 Nov 2014 12:57:08 +0000 (13:57 +0100)]
sd-dhcp-lease: fix copy-paste error asserting wrong function argument

10 years agobusctl: pass error output to stdout (rather than stderr) when generate tree for all...
Lennart Poettering [Wed, 19 Nov 2014 23:00:22 +0000 (00:00 +0100)]
busctl: pass error output to stdout (rather than stderr) when generate tree for all objects

This is a ton more useful when some services fail, since we continue
crawling then and output everything to a pager.

10 years agobusctl: improve output of service creds
Lennart Poettering [Wed, 19 Nov 2014 22:59:26 +0000 (23:59 +0100)]
busctl: improve output of service creds

10 years agobusctl: add new "introspect" verb for introspecting objects
Lennart Poettering [Wed, 19 Nov 2014 20:12:54 +0000 (21:12 +0100)]
busctl: add new "introspect" verb for introspecting objects

10 years agosd-bus: refuse properties that claim to be both writable and constant at the same...
Lennart Poettering [Wed, 19 Nov 2014 19:52:47 +0000 (20:52 +0100)]
sd-bus: refuse properties that claim to be both writable and constant at the same time

10 years agocore: watchdog bus properties cannot be both writable and constant
Lennart Poettering [Wed, 19 Nov 2014 19:52:23 +0000 (20:52 +0100)]
core: watchdog bus properties cannot be both writable and constant

10 years agotests: fix minor memory leak
Lennart Poettering [Wed, 19 Nov 2014 15:43:41 +0000 (16:43 +0100)]
tests: fix minor memory leak

10 years agoset: make set_consume() actually free the allocated string if the string already...
Lennart Poettering [Wed, 19 Nov 2014 15:43:03 +0000 (16:43 +0100)]
set: make set_consume() actually free the allocated string if the string already is in the set

10 years agobusctl: split out introspection parser from tree logic so that we can reuse it for...
Lennart Poettering [Wed, 19 Nov 2014 15:42:21 +0000 (16:42 +0100)]
busctl: split out introspection parser from tree logic so that we can reuse it for a future "busctl introspect" command

10 years agosystemd-logind.service: set Type=notify
Dave Reisner [Wed, 19 Nov 2014 13:13:34 +0000 (08:13 -0500)]
systemd-logind.service: set Type=notify

The code already calls sd_notify("READY=1"), so we may as well take
advantage of the startup behavior in the unit. The same was done for
the journal in a87a38c20.

10 years agosd-dhcp-lease: load/save client ID
Dan Williams [Tue, 18 Nov 2014 23:13:12 +0000 (17:13 -0600)]
sd-dhcp-lease: load/save client ID

The lease is usually tied to the client ID, so users of the
lease may want to know what client ID it was acquired with.

10 years agosd-dhcp-client: allow getting/setting the client ID
Dan Williams [Tue, 18 Nov 2014 23:01:20 +0000 (17:01 -0600)]
sd-dhcp-client: allow getting/setting the client ID

The client identifier can be in many different formats, not just
the one that systemd creates from the Ethernet MAC address.  Non-
ethernet interfaces may have different client IDs formats.  Users
may also have custom client IDs that the wish to use to preserve
lease options delivered by servers configured with the existing
client ID.

10 years agoupdate TODO
Lennart Poettering [Tue, 18 Nov 2014 22:35:46 +0000 (23:35 +0100)]
update TODO

10 years agosd-dhcp-lease: expose load/save functions
Tom Gundersen [Tue, 18 Nov 2014 18:10:37 +0000 (19:10 +0100)]
sd-dhcp-lease: expose load/save functions

Based on patch by Dan Williams.

10 years agosd-dhcp-client: fix REBOOT state handling
Dan Williams [Tue, 4 Nov 2014 17:20:43 +0000 (11:20 -0600)]
sd-dhcp-client: fix REBOOT state handling

client->secs wasn't getting set in the REBOOT state, causing
an assertion.  REBOOT should work the same way as INIT, per
RFC 2131:

 secs   2  Filled in by client, seconds elapsed since client
           began address acquisition or renewal process.

REBOOT is necessary because some DHCP servers (eg on
home routers) do not hand back the same IP address unless the
'ciaddr' field is filled with that address, which DISCOVER
cannot do per the RFCs.  This leads to multiple leases
on machine reboot or DHCP client restart.

10 years agonamespace: unchecked return value from library
Susant Sahani [Mon, 17 Nov 2014 10:58:57 +0000 (16:28 +0530)]
namespace: unchecked return value from library

fix:

CID 1237553 (#1 of 6): Unchecked return value from library
(CHECKED_RETURN

CID 1237553 (#3 of 6): Unchecked return value from library
(CHECKED_RETURN)

CID 1237553 (#4 of 6): Unchecked return value from library
(CHECKED_RETURN)

CID 1237553 (#5 of 6): Unchecked return value from library
(CHECKED_RETURN

CID 1237553 (#6 of 6): Unchecked return value from library
(CHECKED_RETURN)

10 years agobus-socket: fix CID#996290 Unchecked return value
Susant Sahani [Tue, 11 Nov 2014 08:06:33 +0000 (13:36 +0530)]
bus-socket: fix CID#996290 Unchecked return value

10 years agoresolve: fix CID#1237549 Unchecked return value
Susant Sahani [Tue, 11 Nov 2014 08:06:32 +0000 (13:36 +0530)]
resolve: fix CID#1237549 Unchecked return value

10 years agolog: unchecked return value from library
Susant Sahani [Tue, 11 Nov 2014 08:06:31 +0000 (13:36 +0530)]
log: unchecked return value from library

fix 1237557 Unchecked return value from library

10 years agoudev: silence TOCTOU warning when creating a directory
Ronny Chevalier [Sun, 9 Nov 2014 14:42:23 +0000 (15:42 +0100)]
udev: silence TOCTOU warning when creating a directory

CID#979416. There is no real race here to fix, but lets make coverity
happy and rework the code.

Note that we still fail if the directory is removed _after_ we ran
mkdir(), so the same race is still there. Coverity is complaining, though.
Rewrite the code to make it happy.

(David: rewrote the commit-message to note that this is not a race. If I'm
        wrong, blame me, not Ronny!)

10 years agopam_systemd: remove unused null check
Thomas Hindoe Paaboel Andersen [Sat, 15 Nov 2014 22:43:09 +0000 (23:43 +0100)]
pam_systemd: remove unused null check

username was already checked with isempty() and cannot be null at
this point.

CID#1237766

10 years agoautogen: drop unnecessary -Wno-gnu for the clang shortcut
Thomas Hindoe Paaboel Andersen [Sat, 15 Nov 2014 20:32:59 +0000 (21:32 +0100)]
autogen: drop unnecessary -Wno-gnu for the clang shortcut

It is no longer necessary after 65e3a2cf7c3b399853dd309f702ca5078b7d16ea
where -Wno-gnu-variable-sized-type-not-at-end is added if we detect
that the compiler is clang.

10 years agobus-proxy: avoid redundant name validity checks
Lennart Poettering [Fri, 14 Nov 2014 19:06:20 +0000 (20:06 +0100)]
bus-proxy: avoid redundant name validity checks

Our API calls check the validity of bus names anyway, hence we don't
have to do this before calling them...

10 years agobus-proxy: properly check for bus name prefixes when enforcing policy
Lennart Poettering [Fri, 14 Nov 2014 19:06:01 +0000 (20:06 +0100)]
bus-proxy: properly check for bus name prefixes when enforcing policy

10 years agobus-proxy: drop broken access check in driver
Lennart Poettering [Fri, 14 Nov 2014 17:47:54 +0000 (18:47 +0100)]
bus-proxy: drop broken access check in driver

The access check call was broken (as it tried to read a service name
from the UpdateActivationEnvironment() method call which doesn't carry
any). Also, it's unnecessary to make any access checks here, as we just
forward the call to PID 1 which should do the access checks necessary.

10 years agokmod-setup: improve for "kdbus" word on the kernel cmdline
Lennart Poettering [Fri, 14 Nov 2014 17:02:30 +0000 (18:02 +0100)]
kmod-setup: improve for "kdbus" word on the kernel cmdline

We really shouldn't check for words with "strstr()"...

10 years agokmod: move #ifdef checks for kmod-setup out of main.c into kmod-setup.c
Lennart Poettering [Fri, 14 Nov 2014 16:58:32 +0000 (17:58 +0100)]
kmod: move #ifdef checks for kmod-setup out of main.c into kmod-setup.c

10 years agoupdate TODO
Lennart Poettering [Fri, 14 Nov 2014 16:55:19 +0000 (17:55 +0100)]
update TODO

10 years agobusctl: introduce busctl "get-property" command for reading and dumping object properties
Lennart Poettering [Fri, 14 Nov 2014 16:20:04 +0000 (17:20 +0100)]
busctl: introduce busctl "get-property" command for reading and dumping object properties

10 years agomount-setup: remove mount_setup_late()
Daniel Mack [Fri, 14 Nov 2014 14:18:56 +0000 (15:18 +0100)]
mount-setup: remove mount_setup_late()

Turns out we can just do kmod_setup() earlier, before we do mount_setup(),
so there's no need for mount_setup_late() anymore. Instead, put kdbusfs in
mount_table[].

10 years agobusctl: various tweaks to "busctl tree" output
Lennart Poettering [Fri, 14 Nov 2014 12:18:33 +0000 (13:18 +0100)]
busctl: various tweaks to "busctl tree" output

10 years agoman: various tweaks for busctl(1) man page
Lennart Poettering [Fri, 14 Nov 2014 12:11:45 +0000 (13:11 +0100)]
man: various tweaks for busctl(1) man page

10 years agobusctl: add new "call" command to invoke methods on a service
Lennart Poettering [Fri, 14 Nov 2014 12:11:10 +0000 (13:11 +0100)]
busctl: add new "call" command to invoke methods on a service

10 years agosd-bus: add sd_bus_message_is_empty() for checking whether a message carries any...
Lennart Poettering [Fri, 14 Nov 2014 12:05:01 +0000 (13:05 +0100)]
sd-bus: add sd_bus_message_is_empty() for checking whether a message carries any body

10 years agoman: hostnamectl understands -M too
Lennart Poettering [Fri, 14 Nov 2014 11:55:18 +0000 (12:55 +0100)]
man: hostnamectl understands -M too

10 years agosd-bus: sync with kdbus upstream (ABI break)
Daniel Mack [Thu, 13 Nov 2014 19:33:03 +0000 (20:33 +0100)]
sd-bus: sync with kdbus upstream (ABI break)

kdbus has seen a larger update than expected lately, most notably with
kdbusfs, a file system to expose the kdbus control files:

 * Each time a file system of this type is mounted, a new kdbus
   domain is created.

 * The layout inside each mount point is the same as before, except
   that domains are not hierarchically nested anymore.

 * Domains are therefore also unnamed now.

 * Unmounting a kdbusfs will automatically also detroy the
   associated domain.

 * Hence, the action of creating a kdbus domain is now as
   privileged as mounting a filesystem.

 * This way, we can get around creating dev nodes for everything,
   which is last but not least something that is not limited by
   20-bit minor numbers.

The kdbus specific bits in nspawn have all been dropped now, as nspawn
can rely on the container OS to set up its own kdbus domain, simply by
mounting a new instance.

A new set of mounts has been added to mount things *after* the kernel
modules have been loaded. For now, only kdbus is in this set, which is
invoked with mount_setup_late().

10 years agobus-proxyd: temporarily disable policy checks again
Daniel Mack [Thu, 13 Nov 2014 10:59:29 +0000 (11:59 +0100)]
bus-proxyd: temporarily disable policy checks again

There are issues to investigate on with policies shipped by some
packages, which we'll address later. Move that topic out of the
way for now to bring sd-bus in sync with upstream kdbus.

10 years agoudev: support ENV{}=="" global property matches
Kay Sievers [Thu, 13 Nov 2014 19:35:06 +0000 (20:35 +0100)]
udev: support ENV{}=="" global property matches

10 years agoudev: move global property handling from libudev to udevd
Kay Sievers [Thu, 13 Nov 2014 12:50:01 +0000 (13:50 +0100)]
udev: move global property handling from libudev to udevd

10 years agoudev: switch to systemd logging functions
Kay Sievers [Thu, 13 Nov 2014 12:11:04 +0000 (13:11 +0100)]
udev: switch to systemd logging functions

10 years agolibudev: we do not log errors from libraries
Kay Sievers [Thu, 13 Nov 2014 10:42:24 +0000 (11:42 +0100)]
libudev: we do not log errors from libraries

10 years agotests: add test-execute
Ronny Chevalier [Tue, 11 Nov 2014 19:05:40 +0000 (20:05 +0100)]
tests: add test-execute

add tests for the following directives:
- WorkingDirectory
- Personality
- IgnoreSIGPIPE
- PrivateTmp
- SystemCallFilter: It makes test/TEST-04-SECCOMP obsolete, so it has
  been removed.
- SystemCallErrorNumber
- User
- Group
- Environment

10 years agomanager: allow test run to catch SIGCHLD events
Ronny Chevalier [Tue, 11 Nov 2014 18:48:27 +0000 (19:48 +0100)]
manager: allow test run to catch SIGCHLD events

Otherwise we cannot know when a service exited

10 years agokeymap: Fix special keys on ThinkPad X60/X61 Tablet
Bastien Nocera [Wed, 12 Nov 2014 22:31:08 +0000 (23:31 +0100)]
keymap: Fix special keys on ThinkPad X60/X61 Tablet

KEY_DIRECTION is mapped to XF86RotateWindows, to rotate the display:
http://cgit.freedesktop.org/xkeyboard-config/commit/symbols/inet?id=ec875f6f9b7c4028e11d32b071989c682e6502bd

And F13 is mapped to XF86Tools, which is closest to the original toolbox
usage:
http://cgit.freedesktop.org/xkeyboard-config/tree/symbols/inet?id=7a2c4bed212ebbcc05f3c959aef659ce7dd31fd8#n221

10 years agokeymap: Add support for IBM ThinkPad X41 Tablet
Bastien Nocera [Wed, 12 Nov 2014 22:30:46 +0000 (23:30 +0100)]
keymap: Add support for IBM ThinkPad X41 Tablet

Scancode taken from:
http://www.thinkwiki.org/wiki/Tablet_Hardware_Buttons#Linux_Support

William Jon McCann provided the DMI match. IBM seems to have
swapped the version and model of the system:
        Manufacturer: IBM
        Product Name: 18666TU
        Version: ThinkPad X41 Tablet

10 years agoman: fix typo
Lennart Poettering [Wed, 12 Nov 2014 22:44:21 +0000 (23:44 +0100)]
man: fix typo

10 years agoUpdate TODO
Lennart Poettering [Wed, 12 Nov 2014 22:44:14 +0000 (23:44 +0100)]
Update TODO

10 years agoudev: use properly the systemd logging functions
Ronny Chevalier [Wed, 12 Nov 2014 20:42:38 +0000 (21:42 +0100)]
udev: use properly the systemd logging functions

10 years agoudev: use the systemd logging functions in udev tools
Kay Sievers [Wed, 12 Nov 2014 17:35:18 +0000 (18:35 +0100)]
udev: use the systemd logging functions in udev tools

10 years agosd-bus: pass attach flags to BUS_CREATOR_INFO
Daniel Mack [Tue, 11 Nov 2014 17:31:11 +0000 (18:31 +0100)]
sd-bus: pass attach flags to BUS_CREATOR_INFO

kdbus learned parsing the attach flags for the KDBUS_CMD_BUS_CREATOR_INFO
ioctl. Bits not set in this mask will not be exported. Set that field to
_KDBUS_ATTACH_ALL for now.

Signed-off-by: Daniel Mack <daniel@zonque.org>
10 years agoupdate TODO
Lennart Poettering [Tue, 11 Nov 2014 17:12:48 +0000 (18:12 +0100)]
update TODO

10 years agosysuser: simplify access mode syncing by introducing helper function for it
Lennart Poettering [Tue, 11 Nov 2014 17:12:40 +0000 (18:12 +0100)]
sysuser: simplify access mode syncing by introducing helper function for it

10 years agobus-proxyd: explicitly address messages to unique and well-known name
Daniel Mack [Thu, 23 Oct 2014 11:06:38 +0000 (13:06 +0200)]
bus-proxyd: explicitly address messages to unique and well-known name

In order to check for matching policy entries at message transfers, we
have to consider the following:

* check the currently owned names of both the sending and the receiving
peer. If the sending peer is connected via kdbus, the currently owned
names are already attached to the message. If it was originated by the
connection we're proxying for, we store the owned names in our own strv
so we can check against them.

* Walk the list of names to check which name would allow the message to
pass, and explicitly use that name as destination of the message. If the
destination is on kdbus, store both the connection's unique name and the
chosen well-known-name in the message. That way, the kernel will make sure
the supplied name is owned by the supplied unique name, at the time of
sending, and return -EREMCHG otherwise.

* Make the policy checks optional by retrieving the bus owner creds, and
when the uid matches the current user's uid and is non-null, don't check
the bus policy.

10 years agobuild-sys: move libsystemd-capability into libsystemd-shared
Kay Sievers [Tue, 11 Nov 2014 15:13:59 +0000 (16:13 +0100)]
build-sys: move libsystemd-capability into libsystemd-shared

10 years agoRevert "bus-proxyd: make policy checks optional"
Michal Schmidt [Tue, 11 Nov 2014 14:36:00 +0000 (15:36 +0100)]
Revert "bus-proxyd: make policy checks optional"

This reverts commit 5bb24cccbce846c0d77e71b70a3be7f4b2ba6c0e.

It does not even compile (unbalanced {)

10 years agobus-proxyd: make policy checks optional
Daniel Mack [Thu, 23 Oct 2014 11:06:38 +0000 (13:06 +0200)]
bus-proxyd: make policy checks optional

Retrieve the bus owner creds, and when the uid matches the current user's
uid and is non-null, don't check the bus policy.

10 years agobus-proxyd: move name list iteration to policy users
Daniel Mack [Thu, 9 Oct 2014 11:26:53 +0000 (13:26 +0200)]
bus-proxyd: move name list iteration to policy users

We need to figure out which of the possible names satisfied the policy,
so we cannot do the iteration in check_policy_item() but have to leave it
to the users.

Test cases amended accordingly.

10 years agobus-proxyd: enforce policy for method calls
Daniel Mack [Wed, 24 Sep 2014 15:50:31 +0000 (17:50 +0200)]
bus-proxyd: enforce policy for method calls