netdev - Re: [PATCH bpf-next v3 2/2] selftests/bpf: Add bpf_program__attach

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date: Tue, 27 Jun 2023 08:40:46 -0600
From: Daniel Xu <dxu@...uu.xyz>
To: Florian Westphal <fw@...len.de>
Cc: bpf@...r.kernel.org, qde@...cy.de, netdev@...r.kernel.org, 
	andrii.nakryiko@...il.com
Subject: Re: [PATCH bpf-next v3 2/2] selftests/bpf: Add
 bpf_program__attach_netfilter helper test

Hi Florian,

On Tue, Jun 27, 2023 at 01:58:39PM +0200, Florian Westphal wrote:
> Call bpf_program__attach_netfilter() with different
> protocol/hook/priority combinations.
> 
> Test fails if supposedly-illegal attachments work
> (e.g., bogus protocol family, illegal priority and so on) or if a
> should-work attachment fails.  Expected output:
> 
>  ./test_progs -t netfilter_link_attach
>  #145/1   netfilter_link_attach/netfilter link attach 0:OK
>  #145/2   netfilter_link_attach/netfilter link attach 1:OK
>  #145/3   netfilter_link_attach/netfilter link attach 2:OK
>  #145/4   netfilter_link_attach/netfilter link attach 3:OK
>  #145/5   netfilter_link_attach/netfilter link attach 4:OK
>  #145/6   netfilter_link_attach/netfilter link attach 5:OK
>  #145/7   netfilter_link_attach/netfilter link attach 6:OK
>  #145/8   netfilter_link_attach/netfilter link attach 7:OK
>  #145/9   netfilter_link_attach/netfilter link attach 8:OK
>  #145     netfilter_link_attach:OK
> 
> Signed-off-by: Florian Westphal <fw@...len.de>
> ---
>  .../bpf/prog_tests/netfilter_link_attach.c    | 88 +++++++++++++++++++
>  .../bpf/progs/test_netfilter_link_attach.c    | 14 +++
>  2 files changed, 102 insertions(+)
>  create mode 100644 tools/testing/selftests/bpf/prog_tests/netfilter_link_attach.c
>  create mode 100644 tools/testing/selftests/bpf/progs/test_netfilter_link_attach.c
> 
> diff --git a/tools/testing/selftests/bpf/prog_tests/netfilter_link_attach.c b/tools/testing/selftests/bpf/prog_tests/netfilter_link_attach.c
> new file mode 100644
> index 000000000000..dfec6c44f81d
> --- /dev/null
> +++ b/tools/testing/selftests/bpf/prog_tests/netfilter_link_attach.c
> @@ -0,0 +1,88 @@
> +// SPDX-License-Identifier: GPL-2.0-or-later
> +
> +#include <netinet/in.h>
> +#include <linux/netfilter.h>
> +
> +#include "test_progs.h"
> +#include "test_netfilter_link_attach.skel.h"
> +
> +struct nf_link_test {
> +	__u32 pf;
> +	__u32 hooknum;
> +	__s32 priority;
> +	__u32 flags;
> +
> +	bool expect_success;
> +};
> +
> +struct nf_link_test nf_hook_link_tests[] = {
> +	{  },
> +	{ .pf = NFPROTO_NUMPROTO, },
> +	{ .pf = NFPROTO_IPV4, .hooknum = 42, },
> +	{ .pf = NFPROTO_IPV4, .priority = INT_MIN },
> +	{ .pf = NFPROTO_IPV4, .priority = INT_MAX },
> +	{ .pf = NFPROTO_IPV4, .flags = UINT_MAX },
> +
> +	{ .pf = NFPROTO_INET, .priority = 1, },
> +
> +	{ .pf = NFPROTO_IPV4, .priority = -10000, .expect_success = true },
> +	{ .pf = NFPROTO_IPV6, .priority = 10001, .expect_success = true },
> +};
> +
> +void test_netfilter_link_attach(void)
> +{
> +	struct test_netfilter_link_attach *skel;
> +	struct bpf_program *prog;
> +	LIBBPF_OPTS(bpf_netfilter_opts, opts);
> +	int i;
> +
> +	skel = test_netfilter_link_attach__open_and_load();
> +	if (!ASSERT_OK_PTR(skel, "test_netfilter_link_attach__open_and_load"))
> +		goto out;
> +
> +	prog = skel->progs.nf_link_attach_test;
> +	if (!ASSERT_OK_PTR(prog, "attach program"))
> +		goto out;
> +
> +	for (i = 0; i < ARRAY_SIZE(nf_hook_link_tests); i++) {
> +		struct bpf_link *link;
> +		char name[128];
> +
> +		snprintf(name, sizeof(name), "netfilter link attach %u", i);
> +
> +		if (!test__start_subtest(name))
> +			continue;

Nit: naming by index makes it a little annoying to debug failures. How
about adding a name field to struct nf_link_test?

[...]

Thanks,
Daniel