On Mon, Sep 25, 2023 at 10:59 PM Daniel Borkmann <daniel@xxxxxxxxxxxxx> wrote:
>
> This adds bpf_program__attach_meta() API to libbpf. Overall it is very
> similar to tcx. The API looks as following:
>
> LIBBPF_API struct bpf_link *
> bpf_program__attach_meta(const struct bpf_program *prog, int ifindex,
> bool peer_device, const struct bpf_meta_opts *opts);
>
> The struct bpf_meta_opts is done in similar way as struct bpf_tcx_opts.
> bpf_program__attach_meta() compared to bpf_program__attach_tcx() has one
> additional argument, that is peer_device. The latter denotes whether the
> program should be attached to the relative peer of ifindex or whether it
> should be attached to ifindex itself.
>
> Signed-off-by: Daniel Borkmann <daniel@xxxxxxxxxxxxx>
> ---
> tools/lib/bpf/bpf.c | 16 +++++++++++
> tools/lib/bpf/bpf.h | 5 ++++
> tools/lib/bpf/libbpf.c | 61 ++++++++++++++++++++++++++++++++++++----
> tools/lib/bpf/libbpf.h | 15 ++++++++++
> tools/lib/bpf/libbpf.map | 1 +
> 5 files changed, 92 insertions(+), 6 deletions(-)
>
> diff --git a/tools/lib/bpf/bpf.c b/tools/lib/bpf/bpf.c
> index b0f1913763a3..f1335333b63c 100644
> --- a/tools/lib/bpf/bpf.c
> +++ b/tools/lib/bpf/bpf.c
> @@ -810,6 +810,22 @@ int bpf_link_create(int prog_fd, int target_fd,
> if (!OPTS_ZEROED(opts, tcx))
> return libbpf_err(-EINVAL);
> break;
> + case BPF_META_PRIMARY:
> + case BPF_META_PEER:
thinking out loud: should this be expected_attach_type during program
load? Or is it going to be common for primary and peer to be exactly
the same instance of a BPF program? If BPF_META_PRIMARY or
BPF_META_PEER is expected_attach_type, it seems to be a more natural
API where you'll be just saying "bpf_program__attach_meta(prog,
ifindex)" and whether it's primary or peer will be determined by SEC()
definition (SEC("meta/primary") vs SEC("meta/peer"))?
> + relative_fd = OPTS_GET(opts, meta.relative_fd, 0);
> + relative_id = OPTS_GET(opts, meta.relative_id, 0);
> + if (relative_fd && relative_id)
> + return libbpf_err(-EINVAL);
> + if (relative_id) {
> + attr.link_create.meta.relative_id = relative_id;
> + attr.link_create.flags |= BPF_F_ID;
> + } else {
> + attr.link_create.meta.relative_fd = relative_fd;
> + }
> + attr.link_create.meta.expected_revision = OPTS_GET(opts, meta.expected_revision, 0);
> + if (!OPTS_ZEROED(opts, meta))
> + return libbpf_err(-EINVAL);
> + break;
> default:
> if (!OPTS_ZEROED(opts, flags))
> return libbpf_err(-EINVAL);
> diff --git a/tools/lib/bpf/bpf.h b/tools/lib/bpf/bpf.h
> index 74c2887cfd24..175cfb95a175 100644
> --- a/tools/lib/bpf/bpf.h
> +++ b/tools/lib/bpf/bpf.h
> @@ -415,6 +415,11 @@ struct bpf_link_create_opts {
> __u32 relative_id;
> __u64 expected_revision;
> } tcx;
> + struct {
> + __u32 relative_fd;
> + __u32 relative_id;
> + __u64 expected_revision;
> + } meta;
> };
> size_t :0;
> };
> diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
> index b4758e54a815..4d4da8ba2179 100644
> --- a/tools/lib/bpf/libbpf.c
> +++ b/tools/lib/bpf/libbpf.c
> @@ -121,6 +121,8 @@ static const char * const attach_type_name[] = {
> [BPF_TCX_INGRESS] = "tcx_ingress",
> [BPF_TCX_EGRESS] = "tcx_egress",
> [BPF_TRACE_UPROBE_MULTI] = "trace_uprobe_multi",
> + [BPF_META_PRIMARY] = "meta",
> + [BPF_META_PEER] = "meta",
> };
>
> static const char * const link_type_name[] = {
> @@ -137,6 +139,7 @@ static const char * const link_type_name[] = {
> [BPF_LINK_TYPE_NETFILTER] = "netfilter",
> [BPF_LINK_TYPE_TCX] = "tcx",
> [BPF_LINK_TYPE_UPROBE_MULTI] = "uprobe_multi",
> + [BPF_LINK_TYPE_META] = "meta",
> };
>
> static const char * const map_type_name[] = {
> @@ -8910,6 +8913,7 @@ static const struct bpf_sec_def section_defs[] = {
> SEC_DEF("tc", SCHED_CLS, 0, SEC_NONE), /* deprecated / legacy, use tcx */
> SEC_DEF("classifier", SCHED_CLS, 0, SEC_NONE), /* deprecated / legacy, use tcx */
> SEC_DEF("action", SCHED_ACT, 0, SEC_NONE), /* deprecated / legacy, use tcx */
> + SEC_DEF("meta", SCHED_CLS, 0, SEC_NONE),
> SEC_DEF("tracepoint+", TRACEPOINT, 0, SEC_NONE, attach_tp),
> SEC_DEF("tp+", TRACEPOINT, 0, SEC_NONE, attach_tp),
> SEC_DEF("raw_tracepoint+", RAW_TRACEPOINT, 0, SEC_NONE, attach_raw_tp),
> @@ -12019,11 +12023,11 @@ static int attach_lsm(const struct bpf_program *prog, long cookie, struct bpf_li
> }
>
> static struct bpf_link *
> -bpf_program_attach_fd(const struct bpf_program *prog,
> - int target_fd, const char *target_name,
> - const struct bpf_link_create_opts *opts)
> +bpf_program_attach_fd_type(const struct bpf_program *prog,
> + int target_fd, const char *target_name,
> + enum bpf_attach_type attach_type,
> + const struct bpf_link_create_opts *opts)
> {
> - enum bpf_attach_type attach_type;
> char errmsg[STRERR_BUFSIZE];
> struct bpf_link *link;
> int prog_fd, link_fd;
> @@ -12038,8 +12042,6 @@ bpf_program_attach_fd(const struct bpf_program *prog,
> if (!link)
> return libbpf_err_ptr(-ENOMEM);
> link->detach = &bpf_link__detach_fd;
> -
> - attach_type = bpf_program__expected_attach_type(prog);
> link_fd = bpf_link_create(prog_fd, target_fd, attach_type, opts);
> if (link_fd < 0) {
> link_fd = -errno;
> @@ -12053,6 +12055,16 @@ bpf_program_attach_fd(const struct bpf_program *prog,
> return link;
> }
>
> +static struct bpf_link *
> +bpf_program_attach_fd(const struct bpf_program *prog,
> + int target_fd, const char *target_name,
> + const struct bpf_link_create_opts *opts)
> +{
> + return bpf_program_attach_fd_type(prog, target_fd, target_name,
> + bpf_program__expected_attach_type(prog),
> + opts);
> +}
> +
> struct bpf_link *
> bpf_program__attach_cgroup(const struct bpf_program *prog, int cgroup_fd)
> {
> @@ -12106,6 +12118,43 @@ bpf_program__attach_tcx(const struct bpf_program *prog, int ifindex,
> return bpf_program_attach_fd(prog, ifindex, "tcx", &link_create_opts);
> }
>
> +struct bpf_link *
> +bpf_program__attach_meta(const struct bpf_program *prog, int ifindex,
> + bool peer_device, const struct bpf_meta_opts *opts)
you mentioned that there are plans to also support cases where there
is no primary-peer. Is that going to be a primary-only setup or will
it be some third option? If the latter, should this `bool peer_device`
be an enum then?
> +{
> + LIBBPF_OPTS(bpf_link_create_opts, link_create_opts);
> + enum bpf_attach_type attach_type;
> + __u32 relative_id;
> + int relative_fd;
> +
> + if (!OPTS_VALID(opts, bpf_meta_opts))
> + return libbpf_err_ptr(-EINVAL);
> +
> + relative_id = OPTS_GET(opts, relative_id, 0);
> + relative_fd = OPTS_GET(opts, relative_fd, 0);
> + attach_type = peer_device ? BPF_META_PEER : BPF_META_PRIMARY;
> +
> + /* validate we don't have unexpected combinations of non-zero fields */
> + if (!ifindex) {
> + pr_warn("prog '%s': target netdevice ifindex cannot be zero\n",
> + prog->name);
> + return libbpf_err_ptr(-EINVAL);
> + }
> + if (relative_fd && relative_id) {
> + pr_warn("prog '%s': relative_fd and relative_id cannot be set at the same time\n",
> + prog->name);
> + return libbpf_err_ptr(-EINVAL);
> + }
> +
> + link_create_opts.meta.expected_revision = OPTS_GET(opts, expected_revision, 0);
> + link_create_opts.meta.relative_fd = relative_fd;
> + link_create_opts.meta.relative_id = relative_id;
> + link_create_opts.flags = OPTS_GET(opts, flags, 0);
> +
> + return bpf_program_attach_fd_type(prog, ifindex, "meta", attach_type,
> + &link_create_opts);
> +}
> +
> struct bpf_link *bpf_program__attach_freplace(const struct bpf_program *prog,
> int target_fd,
> const char *attach_func_name)
> diff --git a/tools/lib/bpf/libbpf.h b/tools/lib/bpf/libbpf.h
> index 0e52621cba43..827d29cf9a06 100644
> --- a/tools/lib/bpf/libbpf.h
> +++ b/tools/lib/bpf/libbpf.h
> @@ -800,6 +800,21 @@ LIBBPF_API struct bpf_link *
> bpf_program__attach_tcx(const struct bpf_program *prog, int ifindex,
> const struct bpf_tcx_opts *opts);
>
> +struct bpf_meta_opts {
> + /* size of this struct, for forward/backward compatibility */
> + size_t sz;
> + __u32 flags;
> + __u32 relative_fd;
> + __u32 relative_id;
> + __u64 expected_revision;
nit: move flags to be the last, so we don't have that padding before
expected_revision?
> + size_t :0;
> +};
> +#define bpf_meta_opts__last_field expected_revision
> +
> +LIBBPF_API struct bpf_link *
> +bpf_program__attach_meta(const struct bpf_program *prog, int ifindex,
> + bool peer_device, const struct bpf_meta_opts *opts);
> +
> struct bpf_map;
>
> LIBBPF_API struct bpf_link *bpf_map__attach_struct_ops(const struct bpf_map *map);
> diff --git a/tools/lib/bpf/libbpf.map b/tools/lib/bpf/libbpf.map
> index 57712321490f..2dd4fe2cba3d 100644
> --- a/tools/lib/bpf/libbpf.map
> +++ b/tools/lib/bpf/libbpf.map
> @@ -397,6 +397,7 @@ LIBBPF_1.3.0 {
> bpf_obj_pin_opts;
> bpf_object__unpin;
> bpf_prog_detach_opts;
> + bpf_program__attach_meta;
> bpf_program__attach_netfilter;
> bpf_program__attach_tcx;
> bpf_program__attach_uprobe_multi;
> --
> 2.34.1
>
