rescue.service pulls in /bin/plymouth, which doesn't exist on some
distributions (e.g. Arch Linux). Let's mark it as optional, as it's not
even required by the referencing unit and causes unwanted fails in the
integration testsuite.
Fedora Rawhide renamed dbus.service to dbus-daemon.service - that
breaks tests which require working DBus (e.g. TEST-03-JOBS)
Excerpt from the dbus.spec:
The 'dbus' package is only retained for compatibility purposes. It will
eventually be removed and then replaced by 'Provides: dbus' in the
dbus-daemon package. It will then exclusively be used for other packages to
describe their dependency on a system and user bus. It does not pull in any
particular dbus *implementation*, nor any libraries. These should be pulled
in, if required, via explicit dependencies.
When parsing and installing binaries mentioned in Exec*= lines the
5ed0dcf4d5 commit added parsing logic to drop
prefixes, including handling duplicate exclamation marks. But this did not
handle arbitrary combination of multiple prefixes, ie. StartExec=+-/bin/sh was
parsed as -/bin/sh which then would fail to install.
Instead of using egrep and shell replacements, replace both with sed command
that does it all. This sed script extract a group of characters starting with a
/ up to the first space (if any) after the equals sign. This correctly handles
existing non-prefixed, prefixed, multiple-prefixed commands.
About half commands seem to repeat themself, thus sort -u cuts the list of
binaries to install about in half.
To validate change of behaviour both old and new functions were modified to
echo parsed binaries into separate files, and then diffed. The incorrect
-/bin/sh was missing in the new output.
Without this patch tests fail on default Ubuntu installs.
Nested KVM is very flaky as we learnt from our CI. Hence, let's avoid
KVM whenever we detect we are already running inside of KVM.
Maybe one day nested KVM is fixed, at which point we can turn this on
again, but for now let's simply avoid nested KVM, since reliable CI is
more important than quick CI, I guess.
And yes, avoiding KVM for our qemu runs does make things substantially
slower, but I think it's not a complete loss.
Inspired by @evverx' findings in:
https://github.com/systemd/systemd/pull/8701#issuecomment-380213302
This makes the script wait for the newly created partition to
show up before trying to put a filesystem on it, which should
prevent the tests from failing with the following error:
```
New situation:
Disklabel type: dos
Disk identifier: 0x3541a0ec
Device Boot Start End Sectors Size Id Type
/dev/loop6p1 2048 800767 798720 390M 83 Linux
/dev/loop6p2 800768 819199 18432 9M 83 Linux
The partition table has been altered.
Calling ioctl() to re-read partition table.
Syncing disks.
The file /dev/loop6p1 does not exist and no size was specified.
make: *** [setup] Error 1
F: Failed to mkfs -t ext4
Makefile:4: recipe for target 'setup' failed
```
Let's be more restrictive when validating PID files and MAINPID=
messages: don't accept PIDs that make no sense, and if the configuration
source is not trusted, don't accept out-of-cgroup PIDs. A configuratin
source is considered trusted when the PID file is owned by root, or the
message was received from root.
This should lock things down a bit, in case service authors write out
PID files from unprivileged code or use NotifyAccess=all with
unprivileged code. Note that doing so was always problematic, just now
it's a bit less problematic.
When we open the PID file we'll now use the CHASE_SAFE chase_symlinks()
logic, to ensure that we won't follow an unpriviled-owned symlink to a
privileged-owned file thinking this was a valid privileged PID file,
even though it really isn't.
Fixes: #6632
This catches errors like "ninja not found", missing programs etc. early,
instead of silently ignoring them and trying to boot a broken VM.
In install_config_files(), allow some distro specific files to be absent
(such as /etc/sysconfig/init).
