[PATCH bpf-next] libbpf: reject static entry-point BPF programs

[PATCH bpf-next] libbpf: reject static entry-point BPF programs

[Andrii Nakryiko]

Detect use of static entry-point BPF programs (those with SEC() markings) and
emit error message. This is similar to
c1cccec9c636 ("libbpf: Reject static maps") but for BPF programs.

Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
---
 tools/lib/bpf/libbpf.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index 182bd3d3f728..e58f51b24574 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -677,6 +677,11 @@ bpf_object__add_programs(struct bpf_object *obj, Elf_Data *sec_data,
 			return -LIBBPF_ERRNO__FORMAT;
 		}
 
+		if (sec_idx != obj->efile.text_shndx && GELF_ST_BIND(sym.st_info) == STB_LOCAL) {
+			pr_warn("sec '%s': program '%s' is static and not supported\n", sec_name, name);
+			return -ENOTSUP;
+		}
+
 		pr_debug("sec '%s': found program '%s' at insn offset %zu (%zu bytes), code size %zu insns (%zu bytes)\n",
 			 sec_name, name, sec_off / BPF_INSN_SZ, sec_off, prog_sz / BPF_INSN_SZ, prog_sz);
 
-- 
2.30.2

Re: [PATCH bpf-next] libbpf: reject static entry-point BPF programs

[Song Liu]

On Fri, May 14, 2021 at 1:33 PM Andrii Nakryiko <andrii@kernel.org> wrote:
>
> Detect use of static entry-point BPF programs (those with SEC() markings) and
> emit error message. This is similar to
> c1cccec9c636 ("libbpf: Reject static maps") but for BPF programs.
>
> Signed-off-by: Andrii Nakryiko <andrii@kernel.org>

Acked-by: Song Liu <songliubraving@fb.com>

> ---
>  tools/lib/bpf/libbpf.c | 5 +++++
>  1 file changed, 5 insertions(+)
>
> diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
> index 182bd3d3f728..e58f51b24574 100644
> --- a/tools/lib/bpf/libbpf.c
> +++ b/tools/lib/bpf/libbpf.c
> @@ -677,6 +677,11 @@ bpf_object__add_programs(struct bpf_object *obj, Elf_Data *sec_data,
>                         return -LIBBPF_ERRNO__FORMAT;
>                 }
>
> +               if (sec_idx != obj->efile.text_shndx && GELF_ST_BIND(sym.st_info) == STB_LOCAL) {
> +                       pr_warn("sec '%s': program '%s' is static and not supported\n", sec_name, name);
> +                       return -ENOTSUP;
> +               }
> +
>                 pr_debug("sec '%s': found program '%s' at insn offset %zu (%zu bytes), code size %zu insns (%zu bytes)\n",
>                          sec_name, name, sec_off / BPF_INSN_SZ, sec_off, prog_sz / BPF_INSN_SZ, prog_sz);
>
> --
> 2.30.2
>

Re: [PATCH bpf-next] libbpf: reject static entry-point BPF programs

[Alexei Starovoitov]

On Fri, May 14, 2021 at 1:34 PM Andrii Nakryiko <andrii@kernel.org> wrote:
>
> Detect use of static entry-point BPF programs (those with SEC() markings) and
> emit error message.

Applied. I was wondering whether you've seen such combinations ?

Re: [PATCH bpf-next] libbpf: reject static entry-point BPF programs

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to bpf/bpf-next.git (refs/heads/master):

On Fri, 14 May 2021 12:55:34 -0700 you wrote:
> Detect use of static entry-point BPF programs (those with SEC() markings) and
> emit error message. This is similar to
> c1cccec9c636 ("libbpf: Reject static maps") but for BPF programs.
> 
> Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
> ---
>  tools/lib/bpf/libbpf.c | 5 +++++
>  1 file changed, 5 insertions(+)

Here is the summary with links:
  - [bpf-next] libbpf: reject static entry-point BPF programs
    https://git.kernel.org/bpf/bpf-next/c/513f485ca516

You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html

Re: [PATCH bpf-next] libbpf: reject static entry-point BPF programs

[Andrii Nakryiko]

On Fri, May 14, 2021 at 4:14 PM Alexei Starovoitov
<alexei.starovoitov@gmail.com> wrote:
>
> On Fri, May 14, 2021 at 1:34 PM Andrii Nakryiko <andrii@kernel.org> wrote:
> >
> > Detect use of static entry-point BPF programs (those with SEC() markings) and
> > emit error message.
>
> Applied. I was wondering whether you've seen such combinations ?

Haven't seen this anywhere in the real code, only tested locally by
adding static to one of selftests. Unlikely to break anyone, but good
to be as strict as with maps.