Messages in this thread | | | Date | Tue, 1 Dec 2020 21:54:33 +0100 | From | Jiri Olsa <> | Subject | Re: [RFC v2 1/2] perf: support build BPF skeletons with perf |
| |
On Mon, Nov 30, 2020 at 11:36:46PM -0800, Song Liu wrote: > BPF programs are useful in perf to profile BPF programs. BPF skeleton is > by far the easiest way to write BPF tools. Enable building BPF skeletons > in util/bpf_skel. A dummy bpf skeleton is added. More bpf skeletons will > be added for different use cases. > > Signed-off-by: Song Liu <songliubraving@fb.com> > --- > tools/bpf/bpftool/Makefile | 2 ++ > tools/build/Makefile.feature | 3 +- > tools/perf/Makefile.config | 9 +++++ > tools/perf/Makefile.perf | 52 ++++++++++++++++++++++++++-- > tools/perf/util/bpf_skel/.gitignore | 3 ++ > tools/perf/util/bpf_skel/dummy.bpf.c | 19 ++++++++++ > tools/scripts/Makefile.include | 1 + > 7 files changed, 86 insertions(+), 3 deletions(-) > create mode 100644 tools/perf/util/bpf_skel/.gitignore > create mode 100644 tools/perf/util/bpf_skel/dummy.bpf.c > > diff --git a/tools/bpf/bpftool/Makefile b/tools/bpf/bpftool/Makefile > index f60e6ad3a1dff..a01407ec78dc5 100644 > --- a/tools/bpf/bpftool/Makefile > +++ b/tools/bpf/bpftool/Makefile > @@ -120,6 +120,8 @@ endif > > BPFTOOL_BOOTSTRAP := $(if $(OUTPUT),$(OUTPUT)bpftool-bootstrap,./bpftool-bootstrap) > > +bootstrap: $(BPFTOOL_BOOTSTRAP) > + > BOOTSTRAP_OBJS = $(addprefix $(OUTPUT),main.o common.o json_writer.o gen.o btf.o) > OBJS = $(patsubst %.c,$(OUTPUT)%.o,$(SRCS)) $(OUTPUT)disasm.o > > diff --git a/tools/build/Makefile.feature b/tools/build/Makefile.feature > index 97cbfb31b7625..95a58b5564218 100644 > --- a/tools/build/Makefile.feature > +++ b/tools/build/Makefile.feature > @@ -70,7 +70,8 @@ FEATURE_TESTS_BASIC := \ > libaio \ > libzstd \ > disassembler-four-args \ > - file-handle > + file-handle \ > + clang-bpf-co-re
this needs to go under FEATURE_TESTS_EXTRA and you need to call $(call feature_check,clang-bpf-co-re) before checking on $(feature-clang-bpf-co-re)
otherwise the test-all will mark it as detected even if it's not
> > # FEATURE_TESTS_BASIC + FEATURE_TESTS_EXTRA is the complete list > # of all feature tests > diff --git a/tools/perf/Makefile.config b/tools/perf/Makefile.config > index ce8516e4de34f..cb0cf06e0bb43 100644 > --- a/tools/perf/Makefile.config > +++ b/tools/perf/Makefile.config > @@ -621,6 +621,15 @@ ifndef NO_LIBBPF > endif > endif > > +ifeq ($(feature-clang-bpf-co-re), 0) > + BUILD_BPF_SKEL := 0 > +endif > + > +ifeq ($(BUILD_BPF_SKEL), 1) > + $(call detected,CONFIG_PERF_BPF_SKEL) > + CFLAGS += -DBUILD_BPF_SKEL > +endif > + > dwarf-post-unwind := 1 > dwarf-post-unwind-text := BUG > > diff --git a/tools/perf/Makefile.perf b/tools/perf/Makefile.perf > index 7ce3f2e8b9c74..37b7ffe1db27c 100644 > --- a/tools/perf/Makefile.perf > +++ b/tools/perf/Makefile.perf > @@ -126,6 +126,8 @@ include ../scripts/utilities.mak > # > # Define NO_LIBDEBUGINFOD if you do not want support debuginfod > # > +# Define BUILD_BPF_SKEL to enable BPF skeletons > +# > > # As per kernel Makefile, avoid funny character set dependencies > unexport LC_ALL > @@ -735,7 +737,8 @@ prepare: $(OUTPUT)PERF-VERSION-FILE $(OUTPUT)common-cmds.h archheaders $(drm_ioc > $(x86_arch_prctl_code_array) \ > $(rename_flags_array) \ > $(arch_errno_name_array) \ > - $(sync_file_range_arrays) > + $(sync_file_range_arrays) \ > + bpf-skel
I think the 'prepare' target is misused already with other stuff, there's generated bpf_counter.c dependency on util/bpf_skel/bpf_prog_profiler.skel.h in util/.bpf_counter.o.cmd, that should triger the build no?
we have few other dependencies like that for flex/bison generation perhaps we could follow also that
> > $(OUTPUT)%.o: %.c prepare FORCE > $(Q)$(MAKE) -f $(srctree)/tools/build/Makefile.build dir=$(build-dir) $@ > @@ -1008,7 +1011,52 @@ config-clean: > python-clean: > $(python-clean) > > -clean:: $(LIBTRACEEVENT)-clean $(LIBAPI)-clean $(LIBBPF)-clean $(LIBSUBCMD)-clean $(LIBPERF)-clean config-clean fixdep-clean python-clean > +SKEL_OUT := $(abspath util/bpf_skel) > +SKEL_TMP_OUT := $(abspath util/bpf_skel/.tmp) > +SKELETONS := $(SKEL_OUT)/dummy.skel.h > + > +ifdef BUILD_BPF_SKEL > +CLANG ?= clang > +LLVM_STRIP ?= llvm-strip
please move this up where we have similar setup like HOSTCC and others
> +BPFTOOL_CFLAGS := $(filter-out -D_GNU_SOURCE,$(CFLAGS)) > +BPFTOOL := $(SKEL_TMP_OUT)/bpftool-bootstrap > +LIBBPF_SRC := $(abspath ../lib/bpf) > +BPFOBJ := $(SKEL_TMP_OUT)/libbpf.a
we already build libbpf.a, chekc 'LIBBPF = $(BPF_PATH)libbpf.a' above
> +BPF_INCLUDE := $(SKEL_TMP_OUT) > +submake_extras := feature_display=0 > + > +$(SKEL_TMP_OUT): > + $(Q)$(MKDIR) -p $@ > + > +$(BPFTOOL): | $(SKEL_TMP_OUT) > + CFLAGS= $(MAKE) $(submake_extras) -C ../bpf/bpftool \ > + OUTPUT=$(SKEL_TMP_OUT)/ bootstrap > + > +$(SKEL_TMP_OUT)/%.bpf.o: util/bpf_skel/%.bpf.c $(BPFOBJ) | $(SKEL_TMP_OUT) > + $(call QUIET_CLANG, $@) > + $(Q)$(CLANG) -g -O2 -target bpf -c $(filter util/bpf_skel/%.bpf.c,$^) -o $@ && \ > + $(LLVM_STRIP) -g $@ > + > +$(SKEL_OUT)/%.skel.h: $(SKEL_TMP_OUT)/%.bpf.o | $(BPFTOOL) > + $(call QUIET_GENSKEL, $@)
is there a reason to use call in here? you could define QUIET_GENSKEL to use $@ and use it the same way as we use the rest of QUIET_* stuf in Makefile.perf
> new file mode 100644 > index 0000000000000..5263e9e6c5d83 > --- /dev/null > +++ b/tools/perf/util/bpf_skel/.gitignore > @@ -0,0 +1,3 @@ > +# SPDX-License-Identifier: GPL-2.0-only > +.tmp > +*.skel.h > \ No newline at end of file > diff --git a/tools/perf/util/bpf_skel/dummy.bpf.c b/tools/perf/util/bpf_skel/dummy.bpf.c > new file mode 100644 > index 0000000000000..492a43a851deb > --- /dev/null > +++ b/tools/perf/util/bpf_skel/dummy.bpf.c
hum, what's the reason for dummy skeleton? it just adds time to compilation no?
thanks, jirka
| |