unit_write_drop_in{,_private}{,_format} are all affected.
We already append a header to the file (and section markers), so those functions
can only be used to write a whole file at once. Including the newline at
the end feels natural.
After this commit newlines will be duplicated. They will be removed in
subsequent commit.
Also, rewrap the "autogenerated" header to fit within 80 columns.
Michael Biebl [Sat, 28 May 2016 09:03:03 +0000 (11:03 +0200)]
login: Fix policy for org.freedesktop.login1.set-self-linger (#3365) (#3373)
We need to explicitly define authorizations for allow_inactive and
allow_active. Otherwise one is getting "Access denied" when run from a
local console:
$ loginctl enable-linger
Could not enable linger: Access denied
Tejun Heo [Fri, 27 May 2016 16:10:18 +0000 (09:10 -0700)]
core: add cgroup memory controller support on the unified hierarchy (#3315)
On the unified hierarchy, memory controller implements three control knobs -
low, high and max which enables more useable and versatile control over memory
usage. This patch implements support for the three control knobs.
* MemoryLow, MemoryHigh and MemoryMax are added for memory.low, memory.high and
memory.max, respectively.
* As all absolute limits on the unified hierarchy use "max" for no limit, make
memory limit parse functions accept "max" in addition to "infinity" and
document "max" for the new knobs.
* Implement compatibility translation between MemoryMax and MemoryLimit.
v2:
- Fixed missing else's in config_parse_memory_limit().
- Fixed missing newline when writing out drop-ins.
- Coding style updates to use "val > 0" instead of "val".
- Minor updates to documentation.
Djalal Harouni [Thu, 26 May 2016 20:42:29 +0000 (22:42 +0200)]
nspawn: split out seccomp call into nspawn-seccomp.[ch]
Split seccomp into nspawn-seccomp.[ch]. Currently there are no changes,
but this will make it easy in the future to share or use the seccomp logic
from systemd core.
Djalal Harouni [Thu, 26 May 2016 10:59:49 +0000 (12:59 +0200)]
nspawn: rename is_procfs_sysfs_or_suchlike() to is_fs_fully_userns_compatible()
Rename is_procfs_sysfs_or_suchlike() to is_fs_fully_userns_compatible()
to give it the real meaning. This may prevent future modifications that
may introduce bugs.
Christian Hesse [Thu, 26 May 2016 13:57:37 +0000 (15:57 +0200)]
{machine,system}ctl: always pass &changes and &n_changes (#3350)
We have to pass addresses of changes and n_changes to
bus_deserialize_and_dump_unit_file_changes(). Otherwise we are hit by
missing information (subsequent calls to unit_file_changes_add() to
not add anything).
Also prevent null pointer dereference in
bus_deserialize_and_dump_unit_file_changes() by asserting.
Let's make the seed actually work as stable seed, and use siphash24 to generate
the series of addresses, instead of the opaque libc random_r().
This not only makes the seed truly work as stable, portable seed, but also
makes the code quite a bit shorter, and removes a couple of memory allocations.
We try to stick to usec_t for encoding time information, do that here too. In
particular, get rid of "int" second specifications, since signed timespans are
a weird thing.
These objects are only useful when multiple threads are involved, as they
operate with atomic operations. Given that our libraries are explicitly not
thread-safe don't make use of RefCnt here, and make things a bit simpler.
assert_return() should only be used to validate user-facing parameters and
state, assert() should be used for checking our own internal state and
parameters.
sd-network: rename "index" field of the various clients to "ifindex"
A field "index" is not particularly precise and also might conflict with libc's
index() function definition. Also, pretty much everywhere else we call this
concept "ifindex", including in networkd, the primary user of these libraries.
Hence, let's fix this up and call this "ifindex" everywhere here too.
Werner Fink [Wed, 18 Nov 2015 11:28:30 +0000 (12:28 +0100)]
ask-password: ask for passphrases not only on the first console of /dev/console
but also on all other consoles. This does help on e.g. mainframes
where often a serial console together with other consoles are
used. Even rack based servers attachted to both a serial console
as well as having a virtual console do sometimes miss a connected
monitor.
To be able to ask on all terminal devices of /dev/console the devices
are collected. If more than one device are found, then on each of the
terminals a inquiring task for passphrase is forked and do not return
to the caller.
Every task has its own session and its own controlling terminal.
If one of the tasks does handle a password, the remaining tasks
will be terminated.
Also let contradictory options on the command of
systemd-tty-ask-password-agent fail.
Spwan for each device of the system console /dev/console a own process.
Replace the system call wait() with with system call waitid().
Use SIGTERM instead of SIGHUP to get unresponsive childs down.
Port the collect_consoles() function forward to a pulbic and strv
based function "get_kernel_consoles()" in terminal-util.c and use this
in tty-ask-password-agent.c.
In [1] Michel Dänzer and Daniel Vetter wrote:
>> The scenario you describe isn't possible if the Wayland compositor
>> directly uses the KMS API of /dev/dri/card*, but it may be possible if
>> the Wayland compositor uses the fbdev API of /dev/fb* instead (e.g. if
>> weston uses its fbdev backend).
>
> Yeah, if both weston and your screen grabber uses native fbdev API you can
> now screenshot your desktop. And since fbdev has no concept of "current
> owner of the display hw" like the drm master, I think this is not fixable.
> At least not just in userspace. Also even with native KMS compositors
> fbdev still doesn't have the concept of ownership, which is why it doesn't
> bother clearing it's buffer before KMS takes over. I agree that this
> should be reverted or at least hidden better.
TBH, I think that privilege separation between processes running under the same
UID is tenuous. Even with drm, in common setups any user process can ptrace the
"current owner of the display" and call DROP_MASTER or do whatever. It *is*
possible to prevent that, e.g. by disabling ptrace using yama.ptrace_scope, or
selinux, and so on, but afaik this is not commonly done. E.g. all Fedora
systems pull in elfutils-default-yama-scope.rpm through dependencies which sets
yama.ptrace_scope=0. And even assuming that ptrace was disabled, it is trivial
to modify files on disk, communicate through dbus, etc; there is just to many
ways for a non-sandboxed process to interact maliciously with the display shell
to close them all off. To achieve real protection, some sort of sandboxing
must be implemented, and in that case there is no need to rely on access mode
on the device files, since much more stringent measures have to be implemented
anyway.
The situation is similar for framebuffer devices. It is common to add
framebuffer users to video group to allow them unlimited access to /dev/fb*.
Using uaccess would be better solution in that case. Also, since there is no
"current owner" limitation like in DRM, processes running under the same UID
should be able to access /proc/<pid-of-display-server>/fd/* and gain access to
the devices. Nevertheless, weston implements a suid wrapper to access the
devices and then drop privileges, and this patch would make this daemon
pointless. So if the weston developers feel that this change reduces security,
I prefer to revert it.
Tom Gundersen [Mon, 23 May 2016 23:34:29 +0000 (01:34 +0200)]
sd-device: udev-db - handle properties with empty value (#3330)
The statemachine was unable to parse properties with empty values,
reported in [0].
When reaching the start of the KEY, we would unconditionally read
one more character before starting to look for the end-of-line.
Simply look for the end-of-line from the first character.
Tejun Heo [Mon, 23 May 2016 20:48:46 +0000 (16:48 -0400)]
core: fix missing newlines when writing out drop-ins for cgroup settings
Except for per-device BlockIO, IO and DeviceAllow/Deny settings, all were
missing newline causing the next drop-in to be concatenated at the end of the
line. Fix it.
Susant Sahani [Mon, 23 May 2016 09:13:57 +0000 (14:43 +0530)]
networkd: networkd: ndisc set SO_BINDTODEVICE on socket (#3294)
From the issue #2004 we are receiving packet even if this
packet is not intended for this interface.
This can be reproduced.
lp3s0: Updating address: 2001:db8:1:0:7e7a:91ff:fe6d:ffe2/64 (valid for 1d)
wlp3s0: Updating address: fe80::7e7a:91ff:fe6d:ffe2/64 (valid forever)
NDisc CLIENT: Received RA from non-link-local address ::. Ignoring.
NDisc CLIENT: Received RA on wrong interface: 2 != 6. Ignoring.
NDisc CLIENT: Received RA on wrong interface: 2 != 3. Ignoring.
enp0s25: Updating address: 2001:db8:1:0:2ad2:44ff:fe6a:ae07/64 (valid for 1d)
enp0s25: Updating address: fe80::2ad2:44ff:fe6a:ae07/64 (valid forever)
NDisc CLIENT: Sent Router Solicitation
NDisc CLIENT: Sent Router Solicitation
NDisc CLIENT: Sent Router Solicitation
NDisc CLIENT: Received RA on wrong interface: 3 != 2. Ignoring.
NDisc CLIENT: Received RA on wrong interface: 3 != 6. Ignoring.
NDisc CLIENT: Received RA from non-link-local address ::. Ignoring.
NDisc CLIENT: Received RA on wrong interface: 2 != 6. Ignoring.
NDisc CLIENT: Received RA on wrong interface: 2 != 3. Ignoring.
enp0s25: Updating address: 2001:db8:1:0:2ad2:44ff:fe6a:ae07/64 (valid for 1d)
enp0s25: Updating address: fe80::2ad2:44ff:fe6a:ae07/64 (valid forever)
run: do not try to use reply after freeing it (#3318)
We'd call sd_bus_message_unref and then proceed to use
variables pointing into the reply buffer (fd and char*).
dup the fd and copy the string before destorying the reply.
units: restore ConditionNeesUpdate=/etc in ldconfig.service (#3311)
In order to support stateless systems that support offline /usr updates
properly, let's restore the ConditionNeesUpdate=/etc line that makes sure we
are run when /usr is updated and this update needs to be propagated to the
/etc/ld.so.conf file stored in /etc.
This reverts part of #2859, which snuck this change in, but really shouldn't
have.
Tom Gundersen [Sat, 21 May 2016 21:00:32 +0000 (23:00 +0200)]
libsystemd-network: use recv(..., 0) instead of read(...) (#3317)
According to recv(2) these should be the same, but that is not true.
Passing a buffer of length 0 to read is defined to be a noop according
to read(2), but passing a buffer of length 0 to recv will discard the
pending pacet.
We can easily hit this as we allocate our buffer size depending on
the size of the incoming packet (using FIONREAD). As pointed out in
issue #3299 simply sending an empty UDP packet to the DHCP client
port will trigger a busy loop in networkd as we are polling on the
socket but never discarding the empty packet.
Make the fix for net/if.h fuckup even worse (#3287)
The original conflict is fixed in the kernel in v4.6-rc7-40-g4a91cb61bb,
but now our work-around causes a compilation failure.
Keep the workaround to support 4.5 kernels for now, and layer
more ugliness on top.
resolved: fix accounting of dns serves on a link (#3291)
After a few link up/down events I got this warning:
May 17 22:05:10 laptop systemd-resolved[2983]: Failed to read DNS servers for interface wlp3s0, ignoring: Argument list too long
tomty89 [Fri, 20 May 2016 10:28:30 +0000 (18:28 +0800)]
[networkd-dhcp6] do not call sd_dhcp6_client_start() from dhcp6_request_address()
Starting the DHCP client doesn't seem like dhcp6_request_address()'s responsibility anyway. Whenever it's called, sd_dhcp6_client_start() is unconditionally called outside of it as well. See ndisc_router_handler() and ndisc_handler() in networkd-ndisc.c.
Franck Bui [Thu, 19 May 2016 14:37:04 +0000 (16:37 +0200)]
systemctl: reload configuration when enabling sysv units too (#3297)
After enabling/disabling a unit, the daemon configuration is expected
to be unless '--no-reload' option is passed.
However this is not done when enabling a sysv units. This can lead to
the following scenario:
$ cp /etc/init.d/named /etc/init.d/foo
$ systemctl enable foo
foo.service is not a native service, redirecting to systemd-sysv-install
Executing /usr/lib/systemd/systemd-sysv-install enable foo
$ systemctl start foo
Failed to start foo.service: Unit foo.service failed to load: No such file or directory.
This can also be seen after installing a package providing a sysv
service: the service can't be started unless 'daemon-reload' is called
manually. This shouldn't be needed and this patch will fix this case
too since during package installation, the service is expected to be
enabled/disabled.
tblume [Thu, 19 May 2016 14:35:27 +0000 (16:35 +0200)]
systemctl: restore the no-sync option for legacy halt (#3249)
The sync() call on shutdown had been removed with commit 57371e5829a61e5ee6c9f98404dfc729d6c62608
together with the no-sync option for the shutdown commands.
The sync call was restored in commit 4a3ad39957399c4a30fc472a804e72907ecaa4f9 but the no-sync option
wasn't re-added.
I think we should restore this option at least for the legacy halt command.