On Mon, Mar 30, 2026 at 7:08 PM Jiri Olsa <[email protected]> wrote:
>
> On Sun, Mar 29, 2026 at 09:43:39PM +0900, Hoyeon Lee wrote:
> > Currently, attach_probe covers manual single-kprobe attaches by
> > func_name, but not by raw address. This commit adds address-based
> > single-kprobe attach subtests for the two underlying attach paths,
> > legacy tracefs/debugfs and PMU-based non-legacy. The new subtests
> > resolve SYS_NANOSLEEP_KPROBE_NAME through kallsyms, pass the result
> > through bpf_kprobe_opts.addr, and verify that kprobe and kretprobe are
> > still triggered.
> >
> > Signed-off-by: Hoyeon Lee <[email protected]>
> > ---
> > .../selftests/bpf/prog_tests/attach_probe.c | 49 +++++++++++++++++++
> > 1 file changed, 49 insertions(+)
> >
> > diff --git a/tools/testing/selftests/bpf/prog_tests/attach_probe.c
> > b/tools/testing/selftests/bpf/prog_tests/attach_probe.c
> > index 9e77e5da7097..64f2ed75779d 100644
> > --- a/tools/testing/selftests/bpf/prog_tests/attach_probe.c
> > +++ b/tools/testing/selftests/bpf/prog_tests/attach_probe.c
> > @@ -123,6 +123,51 @@ static void test_attach_probe_manual(enum
> > probe_attach_mode attach_mode)
> > test_attach_probe_manual__destroy(skel);
> > }
> >
> > +/* manual attach address-based kprobe/kretprobe testings */
> > +static void test_attach_kprobe_by_addr(enum probe_attach_mode attach_mode)
> > +{
> > + DECLARE_LIBBPF_OPTS(bpf_kprobe_opts, kprobe_opts);
> > + struct bpf_link *kprobe_link, *kretprobe_link;
> > + struct test_attach_probe_manual *skel;
> > + unsigned long func_addr;
> > +
> > + if (!ASSERT_OK(load_kallsyms(), "load_kallsyms"))
> > + return;
> > +
> > + func_addr = ksym_get_addr(SYS_NANOSLEEP_KPROBE_NAME);
> > + if (!ASSERT_NEQ(func_addr, 0UL, "func_addr"))
> > + return;
> > +
> > + skel = test_attach_probe_manual__open_and_load();
> > + if (!ASSERT_OK_PTR(skel, "skel_kprobe_manual_open_and_load"))
> > + return;
> > +
> > + kprobe_opts.attach_mode = attach_mode;
> > + kprobe_opts.retprobe = false;
> > + kprobe_opts.addr = func_addr;
> > + kprobe_link =
> > bpf_program__attach_kprobe_opts(skel->progs.handle_kprobe,
> > + NULL, &kprobe_opts);
> > + if (!ASSERT_OK_PTR(kprobe_link, "attach_kprobe_by_addr"))
> > + goto cleanup;
> > + skel->links.handle_kprobe = kprobe_link;
>
> we usually use skel->links.handle_kprobe directly, no need to use
> kprobe_link or kretprobe_link
>
Thanks for the review! I'll change that in v3 and assign directly to
skel->links.handle_kprobe and skel->links.handle_kretprobe.
> > +
> > + kprobe_opts.retprobe = true;
> > + kretprobe_link =
> > bpf_program__attach_kprobe_opts(skel->progs.handle_kretprobe,
> > + NULL, &kprobe_opts);
> > + if (!ASSERT_OK_PTR(kretprobe_link, "attach_kretprobe_by_addr"))
> > + goto cleanup;
> > + skel->links.handle_kretprobe = kretprobe_link;
> > +
> > + /* trigger & validate kprobe && kretprobe */
> > + usleep(1);
> > +
> > + ASSERT_EQ(skel->bss->kprobe_res, 1, "check_kprobe_res");
> > + ASSERT_EQ(skel->bss->kretprobe_res, 2, "check_kretprobe_res");
> > +
> > +cleanup:
> > + test_attach_probe_manual__destroy(skel);
> > +}
> > +
> > /* attach uprobe/uretprobe long event name testings */
> > static void test_attach_uprobe_long_event_name(void)
> > {
> > @@ -416,6 +461,10 @@ void test_attach_probe(void)
> > test_attach_probe_manual(PROBE_ATTACH_MODE_PERF);
> > if (test__start_subtest("manual-link"))
> > test_attach_probe_manual(PROBE_ATTACH_MODE_LINK);
> > + if (test__start_subtest("kprobe-legacy-by-addr"))
> > + test_attach_kprobe_by_addr(PROBE_ATTACH_MODE_LEGACY);
> > + if (test__start_subtest("kprobe-perf-by-addr"))
> > + test_attach_kprobe_by_addr(PROBE_ATTACH_MODE_PERF);
>
> should we test PROBE_ATTACH_MODE_LINK mode as well?
>
I covered only LEGACY and PERF here, because they test the two
underlying single-kprobe attach paths, legacy tracefs/debugfs and
PMU-based non-legacy.
I treated LINK as duplicate coverage here. If you think explicit LINK
coverage is still worth having, I can add it in v3.
> jirka
>
