On 19/09/2026 02:01, Collin Funk wrote:
Collin Funk <[email protected]> writes:
Those changes look good.
Do you seee tests/cp/capability.sh fail too? It failed in my podman
container.
I think I have an idea to fix it without skipping the entire test.
This diff lets it pass in the container. However, I'm not sure how I
feel about basing it entirely on the return value. I also don't want to
duplicate that 'id -Z 0 2>&1 | ...' invocation and have another place
for it to go out of sync. WDYT?
diff --git a/tests/cp/capability.sh b/tests/cp/capability.sh
index 2efdfda3f..b5d0bc256 100755
--- a/tests/cp/capability.sh
+++ b/tests/cp/capability.sh
@@ -20,6 +20,7 @@
print_ver_ cp
require_root_
working_umask_or_skip_
+getlimits_
grep '^#define HAVE_CAP 1' $CONFIG_HEADER > /dev/null \
@@ -51,6 +52,11 @@ EOF
compare exp err || fail=1
test "$ret" = 1 || fail=1
returns_ 1 test -f copy1 || fail=1
+elif test "$ret" = 1; then
+ cat <<EOF >exp || framework_failure_
+cp: setting attributes for 'copy1': $ENOTSUP
+EOF
+ compare exp err || fail=1
else
compare /dev/null err || fail=1
test "$ret" = 0 || fail=1
Collin
Oh right I hadn't checked root only tests.
Yes this is the most appropriate change for this case I think.
I've tested it here and makes the failure pass.
thanks!
Padraig
