To make debugging much easier, especially for crashes in tests under
QEMU, let's store the entire coredump bundle in the systemd journal,
which is usually kept around by various CIs. Right now, we usually end
up with a journal, but without the coredump itself, which is pretty
useless.
Since ask_password() (and related calls) already append one char, we
ended up appending two. That's not pretty. Let's fix this, and do it
like in all other cases ask_password() (or an equivalent function) is
called.
Otherwise, changing the default gateway doesn't purge old gateway routes
left on the system during daemon restart. This also fixes removing other
foreign gateway routes that don't match the expected configuration.
Tested:
Changed gateway addresses prior to the patch and they lingered on
the system during each reconfiguration. Applied this patch and
reconfigured gateways and other routes multiple times and it removed
the foreign routes that had gateways that didn't match.
Signed-off-by: William A. Kennington III <william@wkennington.com>
This removes a special case that was implemented before: if some service
was already enabled, we'd treat it as having higher priority.
From https://bugzilla.redhat.com/show_bug.cgi?id=1735584#c4:
> Setting ntp off and on should give the same result as just setting it
> on. There should be no stickiness (hidden state). It should behave like
> running an ansible role.
>
> The other service might have been enabled because no other was installed at
> the time. If I install a new NTP service with a higher priority, setting ntp
> on should enable and start the new service, and disable all other. Also, if
> for some reason multiple services are enabled, after setting ntp on there
> should be only one enabled to avoid systemd selecting between them randomly
> on the next boot.
The general idea is that for things which can occur repeatedly, like SetNTP
being called with the same argument, we only log at debug level. For things
which change state, like when we realize that a unit that wasn't enabled
before needs to be enabled, we log at info level.
Also, don't return success if there are no units loaded.
Follow-up for 26ded55709.
The commit says,
> Note that with this change sysinit.target (and thus early boot) is NOT
systematically delayed until the entropy pool is initialized,
But the dependency was not dropped.
This was found by David Seifert (@SoapGentoo).
We explicitly create /etc/systemd/user and other parts of the basic directory
tree. I think we should create /etc/systemd/system too. (The alternative would
be to not create those other directories too, but I think it's nice to have
the basic directory structure in place after installation.)
https://bugzilla.redhat.com/show_bug.cgi?id=1737362
The `coproc` implementation seems to be a little bit different in older
bash versions, so the `strace` is sometimes started AFTER `systemctl
daemon-reload`, which causes unexpected fails. Let's help it a little by
sleeping for a bit.
The MPOL_LOCAL constant is not recognized in current strace versions.
Let's match at least the numerical value of this constant until the
strace patch is approved & merged.
Comparisons are done in the normal order (if (need > available), not if (available < need)),
variables have reduced scope and are renamed for clarity.
The only functional change is that if we return -ENAMETOOLONG, we do that
without modifying the options[] array.
I also added an explanatory comment. The use of one offset to point into three
buffers is not obvious.
Coverity (in CID#1402354) says that sname might be accessed at bad offset, but
I cannot see this happening. We check for available space before writing anything.
Instead of checking for the STA_UNSYNC flag in the timex status, check
the maximum error. It is updated by the kernel, increasing at a rate of
500 ppm. The maximum value is 16 seconds, which triggers the STA_UNSYNC
flag.
This follows timedatex and allows timedated to correctly detect a clock
synchronized by chronyd when configured to not synchronize the RTC.
