diff --git a/pkg/sentry/fsimpl/cgroupfs/cgroupfs.go b/pkg/sentry/fsimpl/cgroupfs/cgroupfs.go
index 504f5cadf..e186c351d 100644
--- a/pkg/sentry/fsimpl/cgroupfs/cgroupfs.go
+++ b/pkg/sentry/fsimpl/cgroupfs/cgroupfs.go
@@ -683,6 +683,8 @@ type writableControllerFileImpl interface {
 // +stateify savable
 type controllerFile struct {
 	kernfs.DynamicBytesFile
+	implStatFS
+
 	allowBackgroundAccess bool
 }
 
diff --git a/test/syscalls/linux/cgroup.cc b/test/syscalls/linux/cgroup.cc
index 3b09dfa04..8bd51ad97 100644
--- a/test/syscalls/linux/cgroup.cc
+++ b/test/syscalls/linux/cgroup.cc
@@ -16,7 +16,9 @@
 // which isn't expected to work, or be safe on a general linux system.
 
 #include <limits.h>
+#include <linux/magic.h>
 #include <sys/mount.h>
+#include <sys/statfs.h>
 #include <unistd.h>
 
 #include <cstdint>
@@ -157,6 +159,21 @@ TEST(Cgroup, ProcsAndTasks) {
   }
 }
 
+TEST(Cgroup, Statfs) {
+  SKIP_IF(!CgroupsAvailable());
+
+  Mounter m(ASSERT_NO_ERRNO_AND_VALUE(TempPath::CreateDir()));
+  Cgroup c = ASSERT_NO_ERRNO_AND_VALUE(m.MountCgroupfs(""));
+
+  struct statfs st;
+  EXPECT_THAT(statfs(c.Relpath("cgroup.procs").c_str(), &st),
+              SyscallSucceeds());
+  EXPECT_EQ(st.f_type, CGROUP_SUPER_MAGIC);
+
+  EXPECT_THAT(statfs(c.Relpath(".").c_str(), &st), SyscallSucceeds());
+  EXPECT_EQ(st.f_type, CGROUP_SUPER_MAGIC);
+}
+
 TEST(Cgroup, ControllersMustBeInUniqueHierarchy) {
   SKIP_IF(!CgroupsAvailable());