systemd-sysctl currently fails silently under any of these conditions:
- Missing permission to write a sysctl.
- Invalid sysctl (path doesn't exists).
- Ignore failure flag ('-' in front of the sysctl name).
Because of this behaviour, configuration issues can go unnoticed as
there is no way to detect those unless going through the logs.
--strict option forces systemd-sysctl to fail if a sysctl is invalid or
if permission are insufficient. Errors on sysctl marked as "ignore
failure" will still be ignored.
This imports credentials also via SMBIOS' "OEM vendor string" section,
similar to the existing import logic from fw_cfg.
Functionality-wise this is very similar to the existing fw_cfg logic,
both of which are easily settable on the qemu command line.
Pros and cons of each:
SMBIOS OEM vendor strings:
- pro: fast, because memory mapped
- pro: somewhat VMM independent, at least in theory
- pro: qemu upstream sees this as the future
- pro: no additional kernel module needed
- con: strings only, thus binary data is base64 encoded
fw_cfg:
- pro: has been supported for longer in qemu
- pro: supports binary data
- con: slow, because IO port based
- con: only qemu
- con: requires qemu_fw_cfg.ko kernel module
- con: qemu upstream sees this as legacy
Otherwise the return value of the last command is propagated, which may
cause spurious test failures. E.g., pkill returns 1 if no process
matched, which may be a problem in cleanup session:
cleanup_session() {
...
pkill -u "$(id -u logind-test-user)"
sleep 1
pkill -KILL -u "$(id -u logind-test-user)"
}
If there are no remaining processes when the final pkill runs, it will
return 1 and therefore cleanup_session will return 1 as well.
Otherwise, idle action may be triggered before starting the test user
session.
This also introduce create_session() and cleanup_session() helper
functions.
Fixes#23952.
Reduce the number of iterations in some of the test cases, since they
generate a huge amount of uevents and basically DoS udev (which can't
keep up while being slowed down by ASan). To avoid this, let's reduce
the number of iterations and bump the timeout when running under ASan,
since we're not interested in performance in such cases.
- use test_append_files() to install additional commands
- drop use of expect
- include assert.sh and use assertions at several places
- use timeout command at several places
- always use logind-test-user
- etc
Replace the call to the `end.service` with `systemctl poweroff`, since
it seems to cause issues no matter what `--job-mode=` is used:
```
[ 129.070993] testsuite-21.sh[380]: ++ systemctl start --job-mode=flush end.service
[ 129.154985] testsuite-21.sh[912]: Failed to start end.service: Transaction for end.service/start is destructive (sysinit.target has 'stop' job queued, but 'start' is included in transaction).
[ 129.159636] testsuite-21.sh[912]: See system logs and 'systemctl status end.service' for details.
```
Also, add a "safety net" which bypasses the manager and does the
poweroff directly, since sometimes the D-Bus call performed by
`systemctl` might timeout (as the manager might be still processing data
from the fuzzing):
```
[ 115.776778] sh[894]: + systemctl poweroff --no-block
[ 166.164242] testsuite-21.sh[893]: Failed to start transient service unit: Connection timed out
[ 166.269289] sh[894]: Call to PowerOff failed: Connection timed out
```
Otherwise it oversaturates the journal, which in some cases can't keep
up with the load of messages (due to the performance penalty caused by
sanitizers), and gets killed by a watchdog.
Let's keep the debug logs in the journal, while logging only
testsute-*.sh stdout/stderr to the console (ba7abf7). This should make
the test output log a bit more readable and potentially the tests itself
a bit faster by avoiding console oversaturation.
Also, it should significantly reduce the size of artifacts kept by CIs.
Yu Watanabe
Quentin Deslandes
Lennart Poettering
David Tardon
Michal Sekletar
Frantisek Sumsal
Luca Boccassi