Re: [PATCH v3 1/2] perf build: Properly guard libbpf includes

From: Ian Rogers <irogers@google.com>
To: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Jiri Olsa <olsajiri@gmail.com>,
	Mike Leach <mike.leach@linaro.org>,
	linux-perf-users@vger.kernel.org, linux-kernel@vger.kernel.org,
	bpf@vger.kernel.org, peterz@infradead.org, mingo@redhat.com,
	mark.rutland@arm.com, alexander.shishkin@linux.intel.com,
	namhyung@kernel.org
Subject: Re: [PATCH v3 1/2] perf build: Properly guard libbpf includes
Date: Tue, 10 Jan 2023 12:00:13 -0800	[thread overview]
Message-ID: <CAP-5=fV4wqgpazr0LpUzdPQ3RsKLNs8SXKttyBwvq2dhe9kPCw@mail.gmail.com> (raw)
In-Reply-To: <Y716Nt3c/Lc0Z4P5@kernel.org>

On Tue, Jan 10, 2023 at 6:46 AM Arnaldo Carvalho de Melo
<acme@kernel.org> wrote:
>
> Em Tue, Jan 10, 2023 at 11:31:12AM -0300, Arnaldo Carvalho de Melo escreveu:
> > Em Mon, Jan 09, 2023 at 11:29:51AM -0800, Ian Rogers escreveu:
> > > On Mon, Jan 9, 2023 at 10:37 AM Ian Rogers <irogers@google.com> wrote:
> > > > -int libbpf_register_prog_handler(const char *sec __maybe_unused,
> > > > -                                 enum bpf_prog_type prog_type __maybe_unused,
> > > > -                                 enum bpf_attach_type exp_attach_type
> > > > __maybe_unused,
> > > > -                                 const struct
> > > > libbpf_prog_handler_opts *opts __maybe_unused)
> > > > +static int libbpf_register_prog_handler(const char *sec __maybe_unused,
> > > > +                                       enum bpf_prog_type prog_type
> > > > __maybe_unused,
> > > > +                                       enum bpf_attach_type
> > > > exp_attach_type __maybe_unused,
> > > > +                                       const void *opts __maybe_unused)
> > > > {
> > > >        pr_err("%s: not support, update libbpf\n", __func__);
> > > >        return -ENOTSUP;
> > > > ```
> > > >
> > > > There are some other fixes necessary too. I'll try to write the fuller
> > > > patch but I have no means for testing except for undefining
> > > > HAVE_LIBBPF_BPF_PROGRAM__SET_INSNS.
> > >
> > > So libbpf_prog_handler_opts is missing in the failing build, this
> > > points to a libbpf before 0.8. I'm somewhat concerned that to work
> > > around these linkage problems we're adding runtime errors - we may
> > > build but the functionality is totally crippled. Is it worth
> > > maintaining these broken builds or to just upfront fail the feature
> > > test?
>
> > Probably better to make the feature test disable bpf support while
> > emitting a warning that features such as a, b, and c won't we available.
>
> This would be the one-liner I think is appropriate for v6.2, ok?
>
> diff --git a/tools/perf/Makefile.config b/tools/perf/Makefile.config
> index 9962ae23ab8c5868..5b87846759036f6f 100644
> --- a/tools/perf/Makefile.config
> +++ b/tools/perf/Makefile.config
> @@ -589,6 +589,8 @@ ifndef NO_LIBELF
>            $(call feature_check,libbpf-bpf_program__set_insns)
>            ifeq ($(feature-libbpf-bpf_program__set_insns), 1)
>              CFLAGS += -DHAVE_LIBBPF_BPF_PROGRAM__SET_INSNS
> +          else
> +            dummy := $(error Error: libbpf devel library needs to be >= 0.8.0 to build with LIBBPF_DYNAMIC, update or build statically with the version that comes with the kernel sources);

It is ok. The intent/result should be the same as:
https://lore.kernel.org/lkml/20230109203424.1157561-2-irogers@google.com/
but it doesn't change a file outside of tools/perf. I think the change
above would be preferable for 6.3.

Thanks,
Ian

>            endif
>            $(call feature_check,libbpf-btf__raw_data)
>            ifeq ($(feature-libbpf-btf__raw_data), 1)