On Tue, Aug 16, 2022 at 2:33 PM Hao Luo <haoluo@xxxxxxxxxx> wrote:
>
> On Mon, Aug 15, 2022 at 8:53 PM Andrii Nakryiko <andrii@xxxxxxxxxx> wrote:
> >
> > Make sure that entire libbpf code base is initializing bpf_attr and
> > perf_event_attr with memset(0). Also for bpf_attr make sure we
> > clear and pass to kernel only relevant parts of bpf_attr. bpf_attr is
> > a huge union of independent sub-command attributes, so there is no need
> > to clear and pass entire union bpf_attr, which over time grows quite
> > a lot and for most commands this growth is completely irrelevant.
> >
> > Few cases where we were relying on compiler initialization of BPF UAPI
> > structs (like bpf_prog_info, bpf_map_info, etc) with `= {};` were
> > switched to memset(0) pattern for future-proofing.
> >
> > Signed-off-by: Andrii Nakryiko <andrii@xxxxxxxxxx>
> > ---
>
> Looks good to me. I went over all the functions in this change and
Thanks!
> verified the conversion is correct. There is only one question: I
> noticed, for bpf_prog_load() and probe_memcg_account(), we only cover
> up to fd_array and attach_btf_obj_fd. Should we cover up to the last
> field i.e. core_relo_rec_size?
libbpf never sets those fields, I think it's only used from light
skeleton generated loaded BPF program, so I felt like it's unnecessary
to include that yet
>
> Acked-by: Hao Luo <haoluo@xxxxxxxxxx>
>
>
> > tools/lib/bpf/bpf.c | 173 ++++++++++++++++++++--------------
> > tools/lib/bpf/libbpf.c | 43 ++++++---
> > tools/lib/bpf/netlink.c | 3 +-
> > tools/lib/bpf/skel_internal.h | 10 +-
> > 4 files changed, 138 insertions(+), 91 deletions(-)
> >
[...]
