On Mon, Feb 7, 2022 at 6:21 AM Alan Maguire <alan.maguire@xxxxxxxxxx> wrote:
>
> On Sat, 5 Feb 2022, Andrii Nakryiko wrote:
>
> > Add a selftest validating various aspects of libbpf's handling of custom
> > SEC() handlers. It also demonstrates how libraries can ensure very early
> > callbacks registration and unregistration using
> > __attribute__((constructor))/__attribute__((destructor)) functions.
> >
> > Signed-off-by: Andrii Nakryiko <andrii@xxxxxxxxxx>
>
> A few suggestions here for additional tests, but
>
> Reviewed-by: Alan Maguire <alan.maguire@xxxxxxxxxx>
>
> Should we override a default attach method to demonstrate that
> custom handlers can do that? Or would that break parallel
> testing mode?
Yep, I should. I was a bit lazy and wanted some feedback before adding
more tests. I'll add some kprobe overload to auto-attach to sys_enter
or something like that. If I do that during the test (not in
constructor/destructor), it won't interfere with parallel tests (they
each run in a different process).
>
> Also might be good to have a test that captured the difference
> in auto-attach behaviour between a skeleton attach and an
> explicit bpf_prog__attach(); running the bpf_prog__attach on the
> SEC("xyz") should result in -EOPNOTSUPP.
Sure, can add that as well.
>
>
> > ---
> > .../bpf/prog_tests/custom_sec_handlers.c | 136 ++++++++++++++++++
> > .../bpf/progs/test_custom_sec_handlers.c | 51 +++++++
> > 2 files changed, 187 insertions(+)
> > create mode 100644 tools/testing/selftests/bpf/prog_tests/custom_sec_handlers.c
> > create mode 100644 tools/testing/selftests/bpf/progs/test_custom_sec_handlers.c
> >
[...]
