Never log when we fail due to OOM when translating enums, let the caller
do that. Translating basic types like enums should be something where
the caller logs, not the translatior functions.
Return -1 when NULL is passed to all enum parser functions.
The non-fallback versions of the enum translator calls already handle
NULL as failure, instead of hitting an assert, and we should do this
here, too.
Let's underline the header line of the table shown by cgtop, how it is
customary for tables. In order to do this, let's introduce new ANSI
underline macros, and clean up the existing ones as side effect.
David Herrmann [Mon, 21 Sep 2015 17:59:07 +0000 (19:59 +0200)]
mount: propagate error codes correctly
Make sure to propagate error codes from mount-loops correctly. Right now,
we return the return-code of the first mount that did _something_. This is
not what we want. Make sure we return an error if _any_ mount fails (and
then make sure to return the first error to not hide proper errors due to
consequential errors like -ENOTDIR).
When Group is set in the unit, the runtime directories are owned by
this group and not the default group of the user (same for cgroup paths
and standard outputs)
Alban Crequy [Fri, 18 Sep 2015 11:37:34 +0000 (13:37 +0200)]
containers: systemd exits with non-zero code
When a systemd service running in a container exits with a non-zero
code, it can be useful to terminate the container immediately and get
the exit code back to the host, when systemd-nspawn returns. This was
not possible to do. This patch adds the following to make it possible:
- Add a read-only "ExitCode" property on PID 1's "Manager" bus object.
By default, it is 0 so the behaviour stays the same as previously.
- Add a method "SetExitCode" on the same object. The method fails when
called on baremetal: it is only allowed in containers or in user
session.
- Add support in systemctl to call "systemctl exit 42". It reuses the
existing code for user session.
- Add exit.target and systemd-exit.service to the system instance.
- Change main() to actually call systemd-shutdown to exit() with the
correct value.
- Add verb 'exit' in systemd-shutdown with parameter --exit-code
- Update systemctl manpage.
Benjamin Robin [Sat, 19 Sep 2015 19:57:51 +0000 (21:57 +0200)]
systemd-notify: Always pass a valid pid to sd_pid_notify
If the option --pid was used, take the pid from this option, unless take
the parend pid. Using 0 as pid (ucred of systemd-notify) will result 99% of the
time in a failure with this error: "Cannot find unit for notify message of PID"
Shouldn't we use always the ppid, since the MAINPID is something else ?
Tom Gundersen [Fri, 21 Aug 2015 21:04:14 +0000 (23:04 +0200)]
sd-ipv4{acd,ll}: add simple test programs
These programs should be run manually, typically two instances on a
veth pair to check conflict detection.
Both test programs take the ifname as input, the ACD also takes the
IP address to check, whereas LL (optionally) takes the seed, which
determines the sequence of IP addresses to try.
Tom Gundersen [Thu, 20 Aug 2015 09:40:10 +0000 (11:40 +0200)]
sd-ipv4ll: rework callbacks
Firstly, no longer distinguish between STOP and INIT states.
Secondly, do not trigger STOP events when calls to sd_ipv4ll_*() fail. The
caller is the one who would receive the event and will already know that the
call to sd_ipv4ll_*() has failed, so it is redundant.
STOP events will now only be triggered by calling sd_ipv4ll_stop() explicitly
or by some internal error in the library triggered by receiving a packet or
an expiring timeout (i.e., any error that would otherwise not be reported
back to the consumer of the library).
Lastly, follow CODING_STYLE and always return NULL on unref. Protect from
objects being destroyed in callbacks accordingly.
Martin Pitt [Thu, 17 Sep 2015 14:09:37 +0000 (16:09 +0200)]
Revert "keymap: Add Corsair K70"
This breaks the same vendor/product ID with the German keyboard layout. As this
is a hack around some weird keyboard driver bug, let's revert until this is
understood better.
See https://github.com/systemd/systemd/issues/1243
Michal Schmidt [Wed, 16 Sep 2015 20:55:02 +0000 (22:55 +0200)]
sd-bus: correct size calculation in DBus fd receive
The size of the allocated array for received file descriptors was
incorrectly calculated. This did not matter when a single file
descriptor was received, but for more descriptors the allocation was
insufficient.
Tom Gundersen [Thu, 3 Sep 2015 23:58:20 +0000 (01:58 +0200)]
resolved: cache - cache what we can of negative redirect chains
When a NXDATA or a NODATA response is received for an alias it may
include CNAME records from the redirect chain. We should cache the
response for each of these names to avoid needless roundtrips in
the future.
It is not sufficient to do the negative caching only for the
canonical name, as the included redirection chain is not guaranteed
to be complete. In fact, only the final CNAME record from the chain
is guaranteed to be included.
We take care not to cache entries that redirects outside the current
zone, as the SOA will then not be valid.
Tom Gundersen [Mon, 17 Aug 2015 05:56:57 +0000 (07:56 +0200)]
resolved: cache - handle CNAME redirection
CNAME records are special in the way they are treated by DNS servers,
and our cache should mimic that behavior: In case a domain name has an
alias, its CNAME record is returned in place of any other.
Our cache was not doing this despite caching the CNAME records, this
entailed needless lookups to re-resolve the CNAME.
Michal Schmidt [Mon, 14 Sep 2015 13:53:47 +0000 (15:53 +0200)]
basic: nicer xsprintf and xstrftime assert messages
It's nicer if the assertion failure message from a bad use of xsprintf
actually mentions xsprintf instead of the expression the macro is
implemented as.
The assert_message_se macro was added in the previous commit as an
internal helper, but it can also be used for customizing assertion
failure messages like in this case.
Example:
char buf[10];
xsprintf(buf, "This is a %s message.\n", "long");
Before:
Assertion '(size_t) snprintf(buf, ELEMENTSOF(buf), "This is a %s
message.\n", "long") < ELEMENTSOF(buf)' failed at foo.c:6, function
main(). Aborting.
After:
Assertion 'xsprintf: buf[] must be big enough' failed at foo.c:6,
function main(). Aborting.
Daniel Mack [Thu, 15 Jan 2015 19:08:42 +0000 (20:08 +0100)]
cgroup: add support for net_cls controllers
Add a new config directive called NetClass= to CGroup enabled units.
Allowed values are positive numbers for fix assignments and "auto" for
picking a free value automatically, for which we need to keep track of
dynamically assigned net class IDs of units. Introduce a hash table for
this, and also record the last ID that was given out, so the allocator
can start its search for the next 'hole' from there. This could
eventually be optimized with something like an irb.
The class IDs up to 65536 are considered reserved and won't be
assigned automatically by systemd. This barrier can be made a config
directive in the future.
Values set in unit files are stored in the CGroupContext of the
unit and considered read-only. The actually assigned number (which
may have been chosen dynamically) is stored in the unit itself and
is guaranteed to remain stable as long as the unit is active.
In the CGroup controller, set the configured CGroup net class to
net_cls.classid. Multiple unit may share the same net class ID,
and those which do are linked together.
Hans de Goede [Mon, 14 Sep 2015 12:16:53 +0000 (14:16 +0200)]
hwdb: Add Thinkpad X1 carbon 3rd gen to 70-pointingstick.hwdb
Like many other recent thinkpads the factory default pointingstick
sensitivity on these devices is quite low, making the pointingstick
very slow in moving the cursor.
This extends the existing hwdb rules for tweaking the sensitivity to
also apply to the X1 carbon 3rd gen model.
cgroup: unify how we invalidate cgroup controller settings
Let's make sure that we follow the same codepaths when adjusting a
cgroup property via the dbus SetProperty() call, and when we execute the
StartupCPUShares= effect.
core: refactor cpu shares/blockio weight cgroup logic
Let's stop using the "unsigned long" type for weights/shares, and let's
just use uint64_t for this, as that's what we expose on the bus.
Unify parsers, and always validate the range for these fields.
Correct the default blockio weight to 500, since that's what the kernel
actually uses.
When parsing the weight/shares settings from unit files accept the empty
string as a way to reset the weight/shares value. When getting it via
the bus, uniformly map (uint64_t) -1 to unset.
Open up StartupCPUShares= and StartupBlockIOWeight= to transient units.