cgroups: fix a leak when using cgroupfs

be sure to load all the existing handlers, so that they can also be
freed in addition to the handlers we treat differently.

Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>

2 files changed, 24 insertions, 0 deletions

diff --git a/pkg/cgroups/cgroups.go b/pkg/cgroups/cgroups.go
index de7f3fd0e..081db772f 100644
--- a/pkg/cgroups/cgroups.go
+++ b/pkg/cgroups/cgroups.go
@@ -328,6 +328,13 @@ func Load(path string) (*CgroupControl, error) {
 		systemd: false,
 	}
 	if !cgroup2 {
+		controllers, err := getAvailableControllers(handlers, false)
+		if err != nil {
+			return nil, err
+		}
+		control.additionalControllers = controllers
+	}
+	if !cgroup2 {
 		for name := range handlers {
 			p := control.getCgroupv1Path(name)
 			if _, err := os.Stat(p); err != nil {
diff --git a/test/e2e/pod_rm_test.go b/test/e2e/pod_rm_test.go
index 0d3f47f30..f0689f152 100644
--- a/test/e2e/pod_rm_test.go
+++ b/test/e2e/pod_rm_test.go
@@ -3,6 +3,8 @@ package integration
 import (
 	"fmt"
 	"os"
+	"path/filepath"
+	"strings"
 
 	. "github.com/containers/libpod/test/utils"
 	. "github.com/onsi/ginkgo"
@@ -40,6 +42,21 @@ var _ = Describe("Podman pod rm", func() {
 		result := podmanTest.Podman([]string{"pod", "rm", podid})
 		result.WaitWithDefaultTimeout()
 		Expect(result.ExitCode()).To(Equal(0))
+
+		// Also check that we don't leak cgroups
+		err := filepath.Walk("/sys/fs/cgroup", func(path string, info os.FileInfo, err error) error {
+			if err != nil {
+				return err
+			}
+			if !info.IsDir() {
+				Expect(err).To(BeNil())
+			}
+			if strings.Contains(info.Name(), podid) {
+				return fmt.Errorf("leaking cgroup path %s", path)
+			}
+			return nil
+		})
+		Expect(err).To(BeNil())
 	})
 
 	It("podman pod rm latest pod", func() {