Re: [PATCH v5 bpf-next 3/3] selftests/bpf: add raw_tp_test_run

On Sep 25, 2020, at 10:31 AM, Andrii Nakryiko [off-list ref] wrote:

On Thu, Sep 24, 2020 at 4:03 PM Song Liu [off-list ref] wrote:

quoted

This test runs test_run for raw_tracepoint program. The test covers ctx
input, retval output, and running on correct cpu.

Signed-off-by: Song Liu <redacted>
---

Few suggestions below, but overall looks good to me:

Acked-by: Andrii Nakryiko <redacted>

quoted

.../bpf/prog_tests/raw_tp_test_run.c          | 98 +++++++++++++++++++
.../bpf/progs/test_raw_tp_test_run.c          | 24 +++++
2 files changed, 122 insertions(+)
create mode 100644 tools/testing/selftests/bpf/prog_tests/raw_tp_test_run.c
create mode 100644 tools/testing/selftests/bpf/progs/test_raw_tp_test_run.c

[...]

quoted

+
+       err = bpf_prog_test_run_xattr(&test_attr);
+       CHECK(err == 0, "test_run", "should fail for too small ctx\n");
+
+       test_attr.ctx_size_in = sizeof(args);
+       err = bpf_prog_test_run_xattr(&test_attr);
+       CHECK(err < 0, "test_run", "err %d\n", errno);
+       CHECK(test_attr.retval != expected_retval, "check_retval",
+             "expect 0x%x, got 0x%x\n", expected_retval, test_attr.retval);
+
+       for (i = 0; i < nr_online; i++) {
+               if (online[i]) {

if (!online[i])
   continue;

That will reduce nestedness by one level

quoted

+                       DECLARE_LIBBPF_OPTS(bpf_test_run_opts, opts,
+                               .ctx_in = args,
+                               .ctx_size_in = sizeof(args),
+                               .flags = BPF_F_TEST_RUN_ON_CPU,
+                               .retval = 0,
+                               .cpu = i,
+                       );

this declares variable, so should be at the top of the lexical scope

quoted

+
+                       err = bpf_prog_test_run_opts(prog_fd, &opts);
+                       CHECK(err < 0, "test_run_opts", "err %d\n", errno);
+                       CHECK(skel->data->on_cpu != i, "check_on_cpu",
+                             "expect %d got %d\n", i, skel->data->on_cpu);
+                       CHECK(opts.retval != expected_retval,
+                             "check_retval", "expect 0x%x, got 0x%x\n",
+                             expected_retval, opts.retval);
+
+                       if (i == 0) {

I agree that this looks a bit obscure. You can still re-use
DECLARE_LIBBPF_OPTS, just move it outside the loop. And then you can
just modify it in place to adjust to a particular case. And in log
output, we'll see 30+ similar success messages for the else branch,
which is indeed unnecessary.

OK.. 2:1, I will change this in v6. 

Thanks,
Song