Re: [PATCH bpf-next 4/4] bpf/selftests: add simple selftest for bpf_smc_ops
From: Zhu Yanjun <hidden>
Date: 2024-11-25 10:52:47
Also in:
bpf, linux-rdma, linux-s390
On 21.11.24 03:00, D. Wythe wrote:
On 11/3/24 9:01 PM, Zhu Yanjun wrote:quoted
在 2024/10/24 4:42, D. Wythe 写道:quoted
From: "D. Wythe" <alibuda@linux.alibaba.com> This PATCH adds a tiny selftest for bpf_smc_ops, to verify the ability to attach and write access. Follow the steps below to run this test. make -C tools/testing/selftests/bpf cd tools/testing/selftests/bpf sudo ./test_progs -t smcThanks a lot. # ./test_progs -t smc #27/1 bpf_smc/load:OK #27 bpf_smc:OK Summary: 1/1 PASSED, 0 SKIPPED, 0 FAILED The above command is based on several kernel modules. After these dependent kernel modules are loaded, then can run the above command successfully. Zhu YanjunHi, Yanjun This is indeed a problem, a better way may be to create a separate testing directory for SMC, and we are trying to do this.
Got it. In the latest patch series, if a test program in sample/bpf can verify this bpf feature, it is better than a selftest program in the directory tools/testing/selftests/bpf. I delved into this selftest tool. It seems that this selftest tool only makes the basic checks. A test program in sample/bpf can do more. I mean, it is very nice that a selftest tool can make selftest on smc bpf. But it is better that a test program in sample/bpf can make some parameter changes in smc. These parameter changes are mentioned in the previous commits. " As a subsequent enhancement, this patch introduces a new hook for eBPF programs that allows decisions on whether to use SMC or not at runtime, including but not limited to local/remote IP address or ports. In simpler words, this feature allows modifications to syn_smc through eBPF programs before the TCP three-way handshake got established. " Zhu Yanjun
Best wishes, D. Wythequoted
quoted
Results shows: Summary: 1/1 PASSED, 0 SKIPPED, 0 FAILED Signed-off-by: D. Wythe <alibuda@linux.alibaba.com> --- .../selftests/bpf/prog_tests/test_bpf_smc.c | 21 +++++++++++ tools/testing/selftests/bpf/progs/bpf_smc.c | 44 ++++++++++++++++++++++ 2 files changed, 65 insertions(+) create mode 100644 tools/testing/selftests/bpf/prog_tests/test_bpf_smc.c create mode 100644 tools/testing/selftests/bpf/progs/bpf_smc.cdiff --git a/tools/testing/selftests/bpf/prog_tests/test_bpf_smc.cb/tools/testing/selftests/bpf/prog_tests/test_bpf_smc.c new file mode 100644 index 00000000..2299853--- /dev/null +++ b/tools/testing/selftests/bpf/prog_tests/test_bpf_smc.c@@ -0,0 +1,21 @@ +// SPDX-License-Identifier: GPL-2.0 +#include <test_progs.h> + +#include "bpf_smc.skel.h" + +static void load(void) +{ + struct bpf_smc *skel; + + skel = bpf_smc__open_and_load(); + if (!ASSERT_OK_PTR(skel, "bpf_smc__open_and_load")) + return; + + bpf_smc__destroy(skel); +} + +void test_bpf_smc(void) +{ + if (test__start_subtest("load")) + load(); +}diff --git a/tools/testing/selftests/bpf/progs/bpf_smc.cb/tools/testing/selftests/bpf/progs/bpf_smc.c new file mode 100644 index 00000000..ebff477--- /dev/null +++ b/tools/testing/selftests/bpf/progs/bpf_smc.c@@ -0,0 +1,44 @@ +// SPDX-License-Identifier: GPL-2.0 + +#include "vmlinux.h" + +#include <bpf/bpf_helpers.h> +#include <bpf/bpf_tracing.h> + +char _license[] SEC("license") = "GPL"; + +struct smc_bpf_ops_ctx { + struct { + struct tcp_sock *tp; + } set_option; + struct { + const struct tcp_sock *tp; + struct inet_request_sock *ireq; + int smc_ok; + } set_option_cond; +}; + +struct smc_bpf_ops { + void (*set_option)(struct smc_bpf_ops_ctx *ctx); + void (*set_option_cond)(struct smc_bpf_ops_ctx *ctx); +}; + +SEC("struct_ops/bpf_smc_set_tcp_option_cond") +void BPF_PROG(bpf_smc_set_tcp_option_cond, struct smc_bpf_ops_ctx*arg) +{ + arg->set_option_cond.smc_ok = 1; +} + +SEC("struct_ops/bpf_smc_set_tcp_option") +void BPF_PROG(bpf_smc_set_tcp_option, struct smc_bpf_ops_ctx *arg) +{ + struct tcp_sock *tp = arg->set_option.tp; + + tp->syn_smc = 1; +} + +SEC(".struct_ops.link") +struct smc_bpf_ops sample_smc_bpf_ops = { + .set_option = (void *) bpf_smc_set_tcp_option, + .set_option_cond = (void *) bpf_smc_set_tcp_option_cond, +};
-- Best Regards, Yanjun.Zhu