Re: [PATCH bpf-next] libbpf: deprecate xdp_cpumap and xdp_devmap sec definitions

[Lorenzo Bianconi <lorenzo@kernel.org>]

[-- Attachment #1: Type: text/plain, Size: 4978 bytes --]

> On Thu, Jan 27, 2022 at 7:37 AM Toke Høiland-Jørgensen <toke@redhat.com> wrote:
> >
> > Lorenzo Bianconi <lorenzo@kernel.org> writes:
> >
> > > Deprecate xdp_cpumap xdp_devmap sec definitions.
> > > Introduce xdp/devmap and xdp/cpumap definitions according to the standard
> > > for SEC("") in libbpf:
> > > - prog_type.prog_flags/attach_place
> > > Update cpumap/devmap samples and kselftests
> > >
> > > Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
> > > ---
> > >  samples/bpf/xdp_redirect_cpu.bpf.c                   |  8 ++++----
> > >  samples/bpf/xdp_redirect_map.bpf.c                   |  2 +-
> > >  samples/bpf/xdp_redirect_map_multi.bpf.c             |  2 +-
> > >  tools/lib/bpf/libbpf.c                               | 12 ++++++++++--
> > >  .../bpf/progs/test_xdp_with_cpumap_frags_helpers.c   |  2 +-
> > >  .../bpf/progs/test_xdp_with_cpumap_helpers.c         |  2 +-
> > >  .../bpf/progs/test_xdp_with_devmap_frags_helpers.c   |  2 +-
> > >  .../bpf/progs/test_xdp_with_devmap_helpers.c         |  2 +-
> > >  .../selftests/bpf/progs/xdp_redirect_multi_kern.c    |  2 +-
> > >  9 files changed, 21 insertions(+), 13 deletions(-)
> > >
> > > diff --git a/samples/bpf/xdp_redirect_cpu.bpf.c b/samples/bpf/xdp_redirect_cpu.bpf.c
> > > index 25e3a405375f..87c54bfdbb70 100644
> > > --- a/samples/bpf/xdp_redirect_cpu.bpf.c
> > > +++ b/samples/bpf/xdp_redirect_cpu.bpf.c
> > > @@ -491,7 +491,7 @@ int  xdp_prognum5_lb_hash_ip_pairs(struct xdp_md *ctx)
> > >       return bpf_redirect_map(&cpu_map, cpu_dest, 0);
> > >  }
> > >
> > > -SEC("xdp_cpumap/redirect")
> > > +SEC("xdp/cpumap")
> > >  int xdp_redirect_cpu_devmap(struct xdp_md *ctx)
> > >  {
> > >       void *data_end = (void *)(long)ctx->data_end;
> > > @@ -507,19 +507,19 @@ int xdp_redirect_cpu_devmap(struct xdp_md *ctx)
> > >       return bpf_redirect_map(&tx_port, 0, 0);
> > >  }
> > >
> > > -SEC("xdp_cpumap/pass")
> > > +SEC("xdp/cpumap")
> > >  int xdp_redirect_cpu_pass(struct xdp_md *ctx)
> > >  {
> > >       return XDP_PASS;
> > >  }
> > >
> > > -SEC("xdp_cpumap/drop")
> > > +SEC("xdp/cpumap")
> > >  int xdp_redirect_cpu_drop(struct xdp_md *ctx)
> > >  {
> > >       return XDP_DROP;
> > >  }
> > >
> > > -SEC("xdp_devmap/egress")
> > > +SEC("xdp/devmap")
> > >  int xdp_redirect_egress_prog(struct xdp_md *ctx)
> > >  {
> > >       void *data_end = (void *)(long)ctx->data_end;
> > > diff --git a/samples/bpf/xdp_redirect_map.bpf.c b/samples/bpf/xdp_redirect_map.bpf.c
> > > index 59efd656e1b2..415bac1758e3 100644
> > > --- a/samples/bpf/xdp_redirect_map.bpf.c
> > > +++ b/samples/bpf/xdp_redirect_map.bpf.c
> > > @@ -68,7 +68,7 @@ int xdp_redirect_map_native(struct xdp_md *ctx)
> > >       return xdp_redirect_map(ctx, &tx_port_native);
> > >  }
> > >
> > > -SEC("xdp_devmap/egress")
> > > +SEC("xdp/devmap")
> > >  int xdp_redirect_map_egress(struct xdp_md *ctx)
> > >  {
> > >       void *data_end = (void *)(long)ctx->data_end;
> > > diff --git a/samples/bpf/xdp_redirect_map_multi.bpf.c b/samples/bpf/xdp_redirect_map_multi.bpf.c
> > > index bb0a5a3bfcf0..8b2fd4ec2c76 100644
> > > --- a/samples/bpf/xdp_redirect_map_multi.bpf.c
> > > +++ b/samples/bpf/xdp_redirect_map_multi.bpf.c
> > > @@ -53,7 +53,7 @@ int xdp_redirect_map_native(struct xdp_md *ctx)
> > >       return xdp_redirect_map(ctx, &forward_map_native);
> > >  }
> > >
> > > -SEC("xdp_devmap/egress")
> > > +SEC("xdp/devmap")
> > >  int xdp_devmap_prog(struct xdp_md *ctx)
> > >  {
> > >       void *data_end = (void *)(long)ctx->data_end;
> > > diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
> > > index 4ce94f4ed34a..1d97bc346be6 100644
> > > --- a/tools/lib/bpf/libbpf.c
> > > +++ b/tools/lib/bpf/libbpf.c
> > > @@ -237,6 +237,8 @@ enum sec_def_flags {
> > >       SEC_SLOPPY_PFX = 16,
> > >       /* BPF program support non-linear XDP buffer */
> > >       SEC_XDP_FRAGS = 32,
> > > +     /* deprecated sec definitions not supposed to be used */
> > > +     SEC_DEPRECATED = 64,
> > >  };
> > >
> > >  struct bpf_sec_def {
> > > @@ -6575,6 +6577,10 @@ static int libbpf_preload_prog(struct bpf_program *prog,
> > >       if (prog->type == BPF_PROG_TYPE_XDP && (def & SEC_XDP_FRAGS))
> > >               opts->prog_flags |= BPF_F_XDP_HAS_FRAGS;
> > >
> > > +     if (def & SEC_DEPRECATED)
> > > +             pr_warn("sec '%s' is deprecated, please use new version instead\n",
> > > +                     prog->sec_name);
> > > +
> >
> > How is the user supposed to figure out what "the new version" is?
> 
> 
> Let's add the section to
> https://github.com/libbpf/libbpf/wiki/Libbpf-1.0-migration-guide and
> link to it from the deprecation warning.

so, is it better to add an utility routine to map the deprecated sec_name to
the new one, or is it enough to add a section in Libbpf-1.0-migration-guide?
I am fine both ways.

Regards,
Lorenzo

> 
> >
> > -Toke
> >

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]