diff options
| author | Lennart Poettering <lennart@poettering.net> | 2018-11-23 01:15:19 +0100 |
|---|---|---|
| committer | Lennart Poettering <lennart@poettering.net> | 2018-11-23 13:41:37 +0100 |
| commit | 43738e001e64ba0ec6522e0c35b67a006abf10e9 (patch) | |
| tree | 82511bf8d14e4bc5ecd8d02b11ff2d7ade0da808 /test/TEST-19-DELEGATE | |
| parent | b8b6f321044ab085358de91a1f72a7d86593dfda (diff) | |
| download | systemd-43738e001e64ba0ec6522e0c35b67a006abf10e9.tar.gz | |
test: extend testcase to ensure controller membership doesn't regress
Diffstat (limited to 'test/TEST-19-DELEGATE')
| -rwxr-xr-x | test/TEST-19-DELEGATE/testsuite.sh | 21 |
1 files changed, 19 insertions, 2 deletions
diff --git a/test/TEST-19-DELEGATE/testsuite.sh b/test/TEST-19-DELEGATE/testsuite.sh index c738bea10e..abfcee3cd2 100755 --- a/test/TEST-19-DELEGATE/testsuite.sh +++ b/test/TEST-19-DELEGATE/testsuite.sh @@ -11,10 +11,27 @@ if grep -q cgroup2 /proc/filesystems ; then -w /sys/fs/cgroup/system.slice/test0.service/cgroup.subtree_control systemd-run --wait --unit=test1.service -p "DynamicUser=1" -p "Delegate=memory pids" \ - grep memory /sys/fs/cgroup/system.slice/test1.service/cgroup.controllers + grep -q memory /sys/fs/cgroup/system.slice/test1.service/cgroup.controllers systemd-run --wait --unit=test2.service -p "DynamicUser=1" -p "Delegate=memory pids" \ - grep pids /sys/fs/cgroup/system.slice/test2.service/cgroup.controllers + grep -q pids /sys/fs/cgroup/system.slice/test2.service/cgroup.controllers + + # "io" is not among the controllers enabled by default for all units, verify that + grep -qv io /sys/fs/cgroup/system.slice/cgroup.controllers + + # Run a service with "io" enabled, and verify it works + systemd-run --wait --unit=test3.service -p "IOAccounting=yes" -p "Slice=system-foo-bar-baz.slice" \ + grep -q io /sys/fs/cgroup/system.slice/system-foo.slice/system-foo-bar.slice/system-foo-bar-baz.slice/test3.service/cgroup.controllers + + # We want to check if "io" is removed again from the controllers + # list. However, PID 1 (rightfully) does this asynchronously. In order + # to force synchronization on this, let's start a short-lived service + # which requires PID 1 to refresh the cgroup tree, so that we can + # verify that this all works. + systemd-run --wait --unit=test4.service true + + # And now check again, "io" should have vanished + grep -qv io /sys/fs/cgroup/system.slice/cgroup.controllers else echo "Skipping TEST-19-DELEGATE, as the kernel doesn't actually support cgroupsv2" >&2 fi |