Re: [PATCH bpf-next v2 2/4] selftests/bpf: convert test_cgroup_storage to test_progs

[Martin KaFai Lau]

On 8/6/24 12:55 AM, Alexis Lothoré (eBPF Foundation) wrote:
> +++ b/tools/testing/selftests/bpf/prog_tests/cgroup_storage.c
> @@ -0,0 +1,65 @@
> +// SPDX-License-Identifier: GPL-2.0
> +
> +#include <test_progs.h>
> +#include "cgroup_helpers.h"
> +#include "cgroup_storage.skel.h"
> +
> +#define TEST_CGROUP "/test-bpf-cgroup-storage-buf/"
> +#define PING_CMD "ping localhost -c 1 -W 1 -q"
> +
> +void test_cgroup_storage(void)
> +{
> +	struct bpf_cgroup_storage_key key;
> +	struct cgroup_storage *skel;
> +	unsigned long long value;
> +	int cgroup_fd;
> +	int err;
> +
> +	cgroup_fd = cgroup_setup_and_join(TEST_CGROUP);

Same. cgroup_fd is leaked.

> +	if (!ASSERT_OK_FD(cgroup_fd, "create cgroup"))
> +		return;
> +
> +	skel = cgroup_storage__open_and_load();
> +	if (!ASSERT_OK_PTR(skel, "load program"))
> +		goto cleanup_cgroup;
> +
> +	skel->links.bpf_prog =
> +		bpf_program__attach_cgroup(skel->progs.bpf_prog, cgroup_fd);
> +	if (!ASSERT_OK_PTR(skel->links.bpf_prog, "attach program"))
> +		goto cleanup_progs;
> +
> +	/* Check that one out of every two packets is dropped */
> +	err = SYS_NOFAIL(PING_CMD);

Better create a netns for network traffic test. It is pretty simple and there 
are existing examples to borrow, e.g. the setup_test_env() in 
prog_tests/sock_addr.c. The netns setup should be done after the 
cgroup_setup_and_join().