libpod: unlock containers when removing pod

It solves a race where a container cleanup process launched because of
the container process exiting normally would hang.

It also solves a problem when running as rootless on cgroup v1 since
it is not possible to force pids.max = 1 on conmon to limit spawning
the cleanup process.

Partially copied from https://github.com/containers/podman/pull/13403

Related to: https://github.com/containers/podman/issues/14057

[NO NEW TESTS NEEDED] it doesn't add any new functionality

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

1 files changed, 13 insertions, 2 deletions

diff --git a/libpod/runtime_pod_linux.go b/libpod/runtime_pod_linux.go
index 2bbccfdf6..62ec7df60 100644
--- a/libpod/runtime_pod_linux.go
+++ b/libpod/runtime_pod_linux.go
@@ -199,10 +199,15 @@ func (r *Runtime) removePod(ctx context.Context, p *Pod, removeCtrs, force bool,
 	// Go through and lock all containers so we can operate on them all at
 	// once.
 	// First loop also checks that we are ready to go ahead and remove.
+	containersLocked := true
 	for _, ctr := range ctrs {
 		ctrLock := ctr.lock
 		ctrLock.Lock()
-		defer ctrLock.Unlock()
+		defer func() {
+			if containersLocked {
+				ctrLock.Unlock()
+			}
+		}()
 
 		// If we're force-removing, no need to check status.
 		if force {
@@ -304,6 +309,12 @@ func (r *Runtime) removePod(ctx context.Context, p *Pod, removeCtrs, force bool,
 		}
 	}
 
+	// let's unlock the containers so if there is any cleanup process, it can terminate its execution
+	for _, ctr := range ctrs {
+		ctr.lock.Unlock()
+	}
+	containersLocked = false
+
 	// Remove pod cgroup, if present
 	if p.state.CgroupPath != "" {
 		logrus.Debugf("Removing pod cgroup %s", p.state.CgroupPath)
@@ -332,7 +343,7 @@ func (r *Runtime) removePod(ctx context.Context, p *Pod, removeCtrs, force bool,
 				}
 			}
 			if err == nil {
-				if err := conmonCgroup.Delete(); err != nil {
+				if err = conmonCgroup.Delete(); err != nil {
 					if removalErr == nil {
 						removalErr = errors.Wrapf(err, "error removing pod %s conmon cgroup", p.ID())
 					} else {