This new unit settings allows restricting which address families are
available to processes. This is an effective way to minimize the attack
surface of services, by turning off entire network stacks for them.
This is based on seccomp, and does not work on x86-32, since seccomp
cannot filter socketcall() syscalls on that platform.
David Herrmann [Tue, 25 Feb 2014 12:08:24 +0000 (13:08 +0100)]
login: set pos-slot to fallback on pos-eviction
If we evict a session position, we open the position slot for future
sessions. However, there might already be another session on the same
position if both were started on the same VT. This is currently done if
gdm spawns the session on its own Xserver.
Hence, look for such a session on pos-eviction and claim the new slot
immediately.
David Herrmann [Tue, 25 Feb 2014 11:20:25 +0000 (12:20 +0100)]
login: fix pos-array allocation
GREEDY_REALLOC takes a pointer to the real size, not the array-width as
argument. Therefore, our array is currently way to small to keep the seat
positions.
Introduce GREEDY_REALLOC0_T() as typed version of GREEDY_REALLOC and store
the array-width instead of array-size.
Łukasz Stelmach [Wed, 19 Feb 2014 15:39:04 +0000 (16:39 +0100)]
smack: relabel directories and files created by systemd
Systemd creates directories in /dev. These directories will
get the label of systemd, which is the label of the System
domain, which is not accessable to everyone. Relabel the
directories, files and symlinks created so that they can be
generally used.
Based on a patch by Casey Schaufler <casey@schaufler-ca.com>.
getty-generator: verify ttys before we make use of them
The ttyS[0-3] devices are weird. They may be enumerated, but when one
actually tries to open and use them they return EIO, because they don't
actually exist. Because they may be enumerated they may be specified on
the kernel command line as console=. And some people do that as default.
As response to that we'll spawn a getty on the tty that will quickly
fail, and we retry a couple of time before giving up. That is quite
noisy.
With this new change we will validate all serial terminals configured
with console= on the kernel cmdline before adding gettys on them, and
remove the invalid ones. THis should remove the noise later on.
Tom Gundersen [Mon, 24 Feb 2014 00:09:21 +0000 (01:09 +0100)]
sd-dhcp-client: respect TP_STATUS_CSUMNOTREADY
If an UDP packet has not passed through a hardware device, its checksum may not
have been computed. This is exposed through the TP_STATUS_CSUMNOTREADY sockopt.
When using raw sockets, skip checksum validation when TP_STATUS_CSUMNOTREADY
is set.
This is necessary for dhcp to work directly over a veth tunnel, e.g. as done
in systemd-nspawn.
core: clean up some confusing regarding SI decimal and IEC binary suffixes for sizes
According to Wikipedia it is customary to specify hardware metrics and
transfer speeds to the basis 1000 (SI decimal), while software metrics
and physical volatile memory (RAM) sizes to the basis 1024 (IEC binary).
So far we specified everything in IEC, let's fix that and be more
true to what's otherwise customary. Since we don't want to parse "Mi"
instead of "M" we document each time what the context used is.
Kay Sievers [Sat, 22 Feb 2014 23:46:26 +0000 (00:46 +0100)]
src/shared/ cannot reference symbols from libraries
../src/shared/unit-name.c:462: error: undefined reference to 'sd_bus_label_escape'
../src/shared/unit-name.c:477: error: undefined reference to 'sd_bus_label_unescape'
collect2: error: ld returned 1 exit status
Tom Gundersen [Sat, 22 Feb 2014 17:23:19 +0000 (18:23 +0100)]
udev: net - allow MTU and Speed to be specified with units
This also changes the names to MTUBytes and BitsPerSecond, respectively. Notice
that the speed was mistakenly documented to be in bytes before this change.
Tom Gundersen [Fri, 21 Feb 2014 21:44:42 +0000 (22:44 +0100)]
networkd/nspawn: enable dhcp client on veth device created in guest
When starting systemd-nspawn with --network-veth, we create a veth device called
host0 in the guest. Pick up on this and start a dhcp client on it. We will also
pick up host0 netdevs created by other containers should they chose to use the
same name.
logind: when we wake up from suspend and the lid is still closed, go to sleep immediately again
This is quite useful on laptops such as the Lenovo Yoga, where the power
button is placed on the front side of the laptop and can be pressed by
accident even if the lid is closed.
This reworks a bit of the logind logic to repeatedly try to suspend the
system as long as a lid is closed. We use the new "post" event source
for this, so that we don't keep things busy.
This also adds some code to check the lid status on boot, so that a
powered-off machine that is accidentaly powered on goes into suspend
immediately.
Yay! From now on I can put my Yoga safely in my backpack without fearing
that it might turn itself on and drain the battery.
This new event source is triggered by the dispatching of any non-post
event source. It can thus be used to do clean-up or recheck work,
triggered by any other event source having been executed.
This is different from "defer" event source which are unconditionally
triggered as long as they are enabled. A "defer" event source that does
nothing will result in the event loop busy looping unless it is turned
off eventually. This is different for "post" event sources that will be
only triggered when some other non-post ran, and will thus not keep the
event loop busy on its own.
It doesn't make any sense to symlink this unit file into /etc when the
unit file itself isn't even installed, with --disable-networkd. This
moves the GENERAL_ALIASES logic into the right "if" block.
Dave Reisner [Fri, 21 Feb 2014 15:31:26 +0000 (10:31 -0500)]
journald: ignore failure to watch hostname_fd on older kernels
Prior to 3.2, /proc/sys/kernel/hostname isn't a pollable file and
sd_event_add_io will return EPERM. Ignore this failure, since it isn't
critical to journald operation.
Apparently bash doesn't turn off non-blocking mode on stdin/stdout when
reading from it, so be nice to bash. Ideally bash would do this on its
own for robustness reasons, though.