Re: [RFC PATCH v2 07/12] selftests/landlock: Add protocol.inval to socket tests

["Günther Noack" <gnoack@google.com>]

On Fri, May 24, 2024 at 05:30:10PM +0800, Mikhail Ivanov wrote:
> Add test that validates behavior of landlock with fully
> access restriction.
> 
> Signed-off-by: Mikhail Ivanov <ivanov.mikhail1@huawei-partners.com>
> ---
> 
> Changes since v1:
> * Refactors commit message.
> ---
>  .../testing/selftests/landlock/socket_test.c  | 34 +++++++++++++++++++
>  1 file changed, 34 insertions(+)
> 
> diff --git a/tools/testing/selftests/landlock/socket_test.c b/tools/testing/selftests/landlock/socket_test.c
> index 31af47de1937..751596c381fe 100644
> --- a/tools/testing/selftests/landlock/socket_test.c
> +++ b/tools/testing/selftests/landlock/socket_test.c
> @@ -265,4 +265,38 @@ TEST_F(protocol, rule_with_unhandled_access)
>  	EXPECT_EQ(0, close(ruleset_fd));
>  }
>  
> +TEST_F(protocol, inval)
> +{
> +	const struct landlock_ruleset_attr ruleset_attr = {
> +		.handled_access_socket = LANDLOCK_ACCESS_SOCKET_CREATE
> +	};
> +
> +	struct landlock_socket_attr protocol = {
> +		.allowed_access = LANDLOCK_ACCESS_SOCKET_CREATE,
> +		.family = self->srv0.protocol.family,
> +		.type = self->srv0.protocol.type,
> +	};
> +
> +	struct landlock_socket_attr protocol_denied = {
> +		.allowed_access = 0,
> +		.family = self->srv0.protocol.family,
> +		.type = self->srv0.protocol.type,
> +	};
> +
> +	int ruleset_fd;
> +
> +	ruleset_fd =
> +		landlock_create_ruleset(&ruleset_attr, sizeof(ruleset_attr), 0);
> +	ASSERT_LE(0, ruleset_fd);
> +
> +	/* Checks zero access value. */
> +	EXPECT_EQ(-1, landlock_add_rule(ruleset_fd, LANDLOCK_RULE_SOCKET,
> +					&protocol_denied, 0));
> +	EXPECT_EQ(ENOMSG, errno);
> +
> +	/* Adds with legitimate values. */
> +	ASSERT_EQ(0, landlock_add_rule(ruleset_fd, LANDLOCK_RULE_SOCKET,
> +				       &protocol, 0));
> +}
> +
>  TEST_HARNESS_MAIN
> -- 
> 2.34.1
> 

Code is based on TEST_F(mini, inval) from net_test.c.  I see that you removed
the check for unhandled allowed_access, because there is already a separate
TEST_F(mini, rule_with_unhandled_access) for that.

That is true for the "legitimate value" case as well, though...?  We already
have a test for that too.  Should that also get removed?

Should we then rename the "inval" test to "rule_with_zero_access", so that the
naming is consistent with the "rule_with_unhandled_access" test?

—Günther