On Thu, Jan 23, 2020 at 7:25 AM KP Singh [off-list ref] wrote:

quoted

From: KP Singh <redacted>

* Add functionality in libbpf to attach eBPF program to LSM hooks
* Lookup the index of the LSM hook in security_hook_heads and pass it in
  attr->lsm_hook_idx

Signed-off-by: KP Singh <redacted>
Reviewed-by: Brendan Jackman <jackmanb@google.com>
Reviewed-by: Florent Revest <redacted>
Reviewed-by: Thomas Garnier <redacted>
---

Looks good, but see few nits below.

Acked-by: Andrii Nakryiko <redacted>

Thanks!

quoted

 tools/lib/bpf/bpf.c      |   6 ++-
 tools/lib/bpf/bpf.h      |   1 +
 tools/lib/bpf/libbpf.c   | 104 +++++++++++++++++++++++++++++++++++++--
 tools/lib/bpf/libbpf.h   |   4 ++
 tools/lib/bpf/libbpf.map |   3 ++
 5 files changed, 114 insertions(+), 4 deletions(-)

[...]

quoted

@@ -5084,6 +5099,8 @@ __bpf_object__open(const char *path, const void *obj_buf, size_t obj_buf_sz,
                if (prog->type != BPF_PROG_TYPE_UNSPEC)
                        continue;

+
+

why these extra lines?

Ah this might have crept in my latest rebase. Will remove these.

quoted

                err = libbpf_prog_type_by_name(prog->section_name, &prog_type,
                                               &attach_type);
                if (err == -ESRCH)

@@ -6160,6 +6177,7 @@ bool bpf_program__is_##NAME(const struct bpf_program *prog)       \
 }                                                              \

 BPF_PROG_TYPE_FNS(socket_filter, BPF_PROG_TYPE_SOCKET_FILTER);
+BPF_PROG_TYPE_FNS(lsm, BPF_PROG_TYPE_LSM);
 BPF_PROG_TYPE_FNS(kprobe, BPF_PROG_TYPE_KPROBE);
 BPF_PROG_TYPE_FNS(sched_cls, BPF_PROG_TYPE_SCHED_CLS);
 BPF_PROG_TYPE_FNS(sched_act, BPF_PROG_TYPE_SCHED_ACT);

@@ -6226,6 +6244,8 @@ static struct bpf_link *attach_raw_tp(const struct bpf_sec_def *sec,
                                      struct bpf_program *prog);
 static struct bpf_link *attach_trace(const struct bpf_sec_def *sec,
                                     struct bpf_program *prog);
+static struct bpf_link *attach_lsm(const struct bpf_sec_def *sec,
+                                  struct bpf_program *prog);

 struct bpf_sec_def {
        const char *sec;

@@ -6272,6 +6292,9 @@ static const struct bpf_sec_def section_defs[] = {
        SEC_DEF("freplace/", EXT,
                .is_attach_btf = true,
                .attach_fn = attach_trace),
+       SEC_DEF("lsm/", LSM,
+               .expected_attach_type = BPF_LSM_MAC,

curious, will there be non-MAC LSM programs? if yes, how they are
going to be different and which prefix will we use then?

One can think BPF_LSM_AUDIT programs which will only be used to log
information from the LSM hooks and not enforce a policy. Currently,
one can sort of do that by disabling CONFIG_SECURITY_BPF_ENFORCE but
that's an all or none hammer.

quoted

+               .attach_fn = attach_lsm),
        BPF_PROG_SEC("xdp",                     BPF_PROG_TYPE_XDP),
        BPF_PROG_SEC("perf_event",              BPF_PROG_TYPE_PERF_EVENT),
        BPF_PROG_SEC("lwt_in",                  BPF_PROG_TYPE_LWT_IN),

@@ -6533,6 +6556,44 @@ static int bpf_object__collect_struct_ops_map_reloc(struct bpf_object *obj,
        return -EINVAL;
 }

+static __s32 find_lsm_hook_idx(struct bpf_program *prog)

nit: I'd stick to int for return result, we barely ever use __s32 in libbpf.c

Sure. Changed to int.

- KP

[...]