[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <40315ef2-e2e0-cb2b-becc-e1ba5a4826ff@isovalent.com>
Date: Wed, 13 Jul 2022 20:03:50 +0100
From: Quentin Monnet <quentin@...valent.com>
To: Andrii Nakryiko <andrii.nakryiko@...il.com>,
Pu Lehui <pulehui@...wei.com>
Cc: bpf <bpf@...r.kernel.org>, Networking <netdev@...r.kernel.org>,
open list <linux-kernel@...r.kernel.org>,
Alexei Starovoitov <ast@...nel.org>,
Daniel Borkmann <daniel@...earbox.net>,
Andrii Nakryiko <andrii@...nel.org>,
Martin KaFai Lau <kafai@...com>,
Song Liu <songliubraving@...com>, Yonghong Song <yhs@...com>,
John Fastabend <john.fastabend@...il.com>,
KP Singh <kpsingh@...nel.org>,
Jean-Philippe Brucker <jean-philippe@...aro.org>
Subject: Re: [PATCH bpf-next 3/3] bpf: iterators: build and use lightweight
bootstrap version of bpftool
On 13/07/2022 19:55, Andrii Nakryiko wrote:
> On Mon, Jul 11, 2022 at 7:37 PM Pu Lehui <pulehui@...wei.com> wrote:
>>
>> kernel/bpf/preload/iterators use bpftool for vmlinux.h, skeleton, and
>> static linking only. So we can use lightweight bootstrap version of
>> bpftool to handle these, and it will be faster.
>>
>> Signed-off-by: Pu Lehui <pulehui@...wei.com>
>> Suggested-by: Andrii Nakryiko <andrii@...nel.org>
>> ---
>> kernel/bpf/preload/iterators/Makefile | 13 +++++++++----
>> 1 file changed, 9 insertions(+), 4 deletions(-)
>>
>> diff --git a/kernel/bpf/preload/iterators/Makefile b/kernel/bpf/preload/iterators/Makefile
>> index bfe24f8c5a20..cf5f39f95fed 100644
>> --- a/kernel/bpf/preload/iterators/Makefile
>> +++ b/kernel/bpf/preload/iterators/Makefile
>> @@ -9,7 +9,7 @@ LLVM_STRIP ?= llvm-strip
>> TOOLS_PATH := $(abspath ../../../../tools)
>> BPFTOOL_SRC := $(TOOLS_PATH)/bpf/bpftool
>> BPFTOOL_OUTPUT := $(abs_out)/bpftool
>> -DEFAULT_BPFTOOL := $(OUTPUT)/sbin/bpftool
>> +DEFAULT_BPFTOOL := $(BPFTOOL_OUTPUT)/bootstrap/bpftool
>> BPFTOOL ?= $(DEFAULT_BPFTOOL)
>>
>> LIBBPF_SRC := $(TOOLS_PATH)/lib/bpf
>> @@ -61,9 +61,14 @@ $(BPFOBJ): $(wildcard $(LIBBPF_SRC)/*.[ch] $(LIBBPF_SRC)/Makefile) | $(LIBBPF_OU
>> OUTPUT=$(abspath $(dir $@))/ prefix= \
>> DESTDIR=$(LIBBPF_DESTDIR) $(abspath $@) install_headers
>>
>> +ifeq ($(CROSS_COMPILE),)
>> $(DEFAULT_BPFTOOL): $(BPFOBJ) | $(BPFTOOL_OUTPUT)
>> $(Q)$(MAKE) $(submake_extras) -C $(BPFTOOL_SRC) \
>> OUTPUT=$(BPFTOOL_OUTPUT)/ \
>> - LIBBPF_OUTPUT=$(LIBBPF_OUTPUT)/ \
>> - LIBBPF_DESTDIR=$(LIBBPF_DESTDIR)/ \
>> - prefix= DESTDIR=$(abs_out)/ install-bin
>> + LIBBPF_BOOTSTRAP_OUTPUT=$(LIBBPF_OUTPUT)/ \
>> + LIBBPF_BOOTSTRAP_DESTDIR=$(LIBBPF_DESTDIR)/ bootstrap
>> +else
>> +$(DEFAULT_BPFTOOL): | $(BPFTOOL_OUTPUT)
>> + $(Q)$(MAKE) $(submake_extras) -C $(BPFTOOL_SRC) \
>> + OUTPUT=$(BPFTOOL_OUTPUT)/ bootstrap
>> +endif
>
> another idea (related to my two previous comments for this patch set),
> maybe we can teach bpftool's Makefile to reuse LIBBPF_OUTPUT as
> LIBBPF_BOOTSTRAP_OUTPUT, if there is no CROSS_COMPILE? Then we can
> keep iterators/Makefile, samples/bpf/Makefile and runqslower/Makefile
> simpler and ignorant of CROSS_COMPILE, but still get the benefit of
> not rebuilding libbpf unnecessarily in non-cross-compile mode?
Could be a good idea. Seeing how the HID BPF patches add BTF/skeletons
generation at new locations, I'm also starting to wonder if it would be
worth having a Makefile.bpftool.include of some sort to harmonise the
way we compile the bootstrap bpftool as a dependency, and make it easier
to maintain. I haven't looked at how feasible that would be, yet.
Powered by blists - more mailing lists