[v3,bpf-next,2/4] libbpf: Add helpers to extract perf fd from bpf_link

It is sometimes necessary to perform ioctl's on the underlying perf fd.
There is not currently a way to extract the fd given a bpf_link, so add a
a pair of casting and getting helpers.

The casting and getting helpers are nice because they let us define
broad categories of links that makes it clear to users what they can
expect to extract from what type of link.

Acked-by: Song Liu <songliubraving@fb.com>
Signed-off-by: Daniel Xu <dxu@dxuuu.xyz>
---
 tools/lib/bpf/libbpf.c   | 21 +++++++++++++++++++++
 tools/lib/bpf/libbpf.h   | 13 +++++++++++++
 tools/lib/bpf/libbpf.map |  4 ++++
 3 files changed, 38 insertions(+)

On Fri, Aug 16, 2019 at 3:32 PM Daniel Xu <dxu@dxuuu.xyz> wrote:
>
> It is sometimes necessary to perform ioctl's on the underlying perf fd.
> There is not currently a way to extract the fd given a bpf_link, so add a
> a pair of casting and getting helpers.
>
> The casting and getting helpers are nice because they let us define
> broad categories of links that makes it clear to users what they can
> expect to extract from what type of link.
>
> Acked-by: Song Liu <songliubraving@fb.com>
> Signed-off-by: Daniel Xu <dxu@dxuuu.xyz>
> ---

This looks great, thanks a lot!

I think you might have a conflict with dadb81d0afe7 ("libbpf: make
libbpf.map source of truth for libbpf version") in libbpf.map, so you
might need to pull, rebase and re-post rebased version. But in any
case:

Acked-by: Andrii Nakryiko <andriin@fb.com>

>  tools/lib/bpf/libbpf.c   | 21 +++++++++++++++++++++
>  tools/lib/bpf/libbpf.h   | 13 +++++++++++++
>  tools/lib/bpf/libbpf.map |  4 ++++
>  3 files changed, 38 insertions(+)
>
> diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
> index 2233f919dd88..41588e13be2b 100644
> --- a/tools/lib/bpf/libbpf.c
> +++ b/tools/lib/bpf/libbpf.c
> @@ -4876,6 +4876,7 @@ int bpf_prog_load_xattr(const struct bpf_prog_load_attr *attr,
>
>  struct bpf_link {
>         int (*destroy)(struct bpf_link *link);
> +       enum bpf_link_type type;
>  };
>
>  int bpf_link__destroy(struct bpf_link *link)
> @@ -4909,6 +4910,24 @@ static int bpf_link__destroy_perf_event(struct bpf_link *link)
>         return err;
>  }
>
> +const struct bpf_link_fd *bpf_link__as_fd(const struct bpf_link *link)
> +{
> +       if (link->type != LIBBPF_LINK_FD)
> +               return NULL;
> +
> +       return (struct bpf_link_fd *)link;
> +}
> +
> +enum bpf_link_type bpf_link__type(const struct bpf_link *link)
> +{
> +       return link->type;
> +}
> +
> +int bpf_link_fd__fd(const struct bpf_link_fd *link)
> +{
> +       return link->fd;
> +}
> +
>  struct bpf_link *bpf_program__attach_perf_event(struct bpf_program *prog,
>                                                 int pfd)
>  {
> @@ -4932,6 +4951,7 @@ struct bpf_link *bpf_program__attach_perf_event(struct bpf_program *prog,
>         if (!link)
>                 return ERR_PTR(-ENOMEM);
>         link->link.destroy = &bpf_link__destroy_perf_event;
> +       link->link.type = LIBBPF_LINK_FD;
>         link->fd = pfd;
>
>         if (ioctl(pfd, PERF_EVENT_IOC_SET_BPF, prog_fd) < 0) {
> @@ -5225,6 +5245,7 @@ struct bpf_link *bpf_program__attach_raw_tracepoint(struct bpf_program *prog,
>         link = malloc(sizeof(*link));
>         if (!link)
>                 return ERR_PTR(-ENOMEM);
> +       link->link.type = LIBBPF_LINK_FD;
>         link->link.destroy = &bpf_link__destroy_fd;
>
>         pfd = bpf_raw_tracepoint_open(tp_name, prog_fd);
> diff --git a/tools/lib/bpf/libbpf.h b/tools/lib/bpf/libbpf.h
> index e8f70977d137..2ddef5315ff9 100644
> --- a/tools/lib/bpf/libbpf.h
> +++ b/tools/lib/bpf/libbpf.h
> @@ -166,7 +166,20 @@ LIBBPF_API int bpf_program__pin(struct bpf_program *prog, const char *path);
>  LIBBPF_API int bpf_program__unpin(struct bpf_program *prog, const char *path);
>  LIBBPF_API void bpf_program__unload(struct bpf_program *prog);
>
> +enum bpf_link_type {
> +       LIBBPF_LINK_FD,
> +};
> +
>  struct bpf_link;
> +struct bpf_link_fd;
> +
> +/* casting APIs */
> +LIBBPF_API const struct bpf_link_fd *
> +bpf_link__as_fd(const struct bpf_link *link);
> +
> +/* getters APIs */
> +LIBBPF_API enum bpf_link_type bpf_link__type(const struct bpf_link *link);
> +LIBBPF_API int bpf_link_fd__fd(const struct bpf_link_fd *link);
>
>  LIBBPF_API int bpf_link__destroy(struct bpf_link *link);
>
> diff --git a/tools/lib/bpf/libbpf.map b/tools/lib/bpf/libbpf.map
> index 4e72df8e98ba..ee9945177100 100644
> --- a/tools/lib/bpf/libbpf.map
> +++ b/tools/lib/bpf/libbpf.map
> @@ -186,4 +186,8 @@ LIBBPF_0.0.4 {
>  } LIBBPF_0.0.3;
>
>  LIBBPF_0.0.5 {
> +       global:
> +               bpf_link__type;
> +               bpf_link__as_fd;
> +               bpf_link_fd__fd;
>  } LIBBPF_0.0.4;
> --
> 2.20.1
>

On Mon, Aug 19, 2019, at 10:45 AM, Andrii Nakryiko wrote:
> On Fri, Aug 16, 2019 at 3:32 PM Daniel Xu <dxu@dxuuu.xyz> wrote:
> >
> > It is sometimes necessary to perform ioctl's on the underlying perf fd.
> > There is not currently a way to extract the fd given a bpf_link, so add a
> > a pair of casting and getting helpers.
> >
> > The casting and getting helpers are nice because they let us define
> > broad categories of links that makes it clear to users what they can
> > expect to extract from what type of link.
> >
> > Acked-by: Song Liu <songliubraving@fb.com>
> > Signed-off-by: Daniel Xu <dxu@dxuuu.xyz>
> > ---
> 
> This looks great, thanks a lot!
> 
> I think you might have a conflict with dadb81d0afe7 ("libbpf: make
> libbpf.map source of truth for libbpf version") in libbpf.map, so you
> might need to pull, rebase and re-post rebased version. But in any
> case:
> 

The patchset is already rebased on top :). Thanks for the tip.

Daniel