Re: [PATCH bpf-next 2/4] selftests/bpf: Add tests for bpf_call_rcu()

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] Incomplete test validation in `test_call_rcu_teardown`.
--

commit 7d7060669166bdbb9948b95d35813ccc322df204
Author: Puranjay Mohan <puranjay@kernel.org>

selftests/bpf: Add tests for bpf_call_rcu()

This commit adds tests for bpf_call_rcu() to cover the callback running
after a grace period, checking -EBUSY on second arm, head reuse, and
teardown. It also includes negative tests for invalid map types and
verifier rejections.

> diff --git a/tools/testing/selftests/bpf/prog_tests/call_rcu.c b/tools/testing/selftests/bpf/prog_tests/call_rcu.c
> new file mode 100644
> index 0000000000000..ccbfa9965d511
> --- /dev/null
> +++ b/tools/testing/selftests/bpf/prog_tests/call_rcu.c

[ ... ]

> +/* Tear down the map and the program while a callback is still queued. */
> +static void test_call_rcu_teardown(void)
> +{
> +	LIBBPF_OPTS(bpf_test_run_opts, opts);
> +	struct call_rcu *skel;
> +
> +	skel = call_rcu__open_and_load();
> +	if (!ASSERT_OK_PTR(skel, "skel_open_and_load"))
> +		return;
> +
> +	/* The callback re-arms after the last user reference is gone; that must be refused. */
> +	skel->bss->chain = 1;
> +	ASSERT_OK(bpf_prog_test_run_opts(bpf_program__fd(skel->progs.arm), &opts), "test_run");
> +	call_rcu__destroy(skel);
> +}

[Severity: Medium]
Does this test actually verify that the re-arming was refused?

It looks like we set chain to 1, run the program, and then immediately
destroy the skeleton without waiting for the callback to run or asserting
the value of chain_err.

If a kernel regression allowed the re-arm to succeed, would this test just
silently pass without catching the failure?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260907134552.1772405-1-puranjay@kernel.org?part=2