[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <87bkj1p8y0.fsf@oracle.com>
Date: Wed, 03 May 2023 20:27:03 +0200
From: "Jose E. Marchesi" <jose.marchesi@...cle.com>
To: Andrii Nakryiko <andrii.nakryiko@...il.com>
Cc: Daniel Rosenberg <drosen@...gle.com>, bpf@...r.kernel.org,
Alexei Starovoitov <ast@...nel.org>,
Daniel Borkmann <daniel@...earbox.net>,
John Fastabend <john.fastabend@...il.com>,
Andrii Nakryiko <andrii@...nel.org>,
Martin KaFai Lau <martin.lau@...ux.dev>,
Song Liu <song@...nel.org>, Yonghong Song <yhs@...com>,
KP Singh <kpsingh@...nel.org>,
Stanislav Fomichev <sdf@...gle.com>,
Hao Luo <haoluo@...gle.com>, Jiri Olsa <jolsa@...nel.org>,
Shuah Khan <shuah@...nel.org>,
Jonathan Corbet <corbet@....net>,
Joanne Koong <joannelkoong@...il.com>,
Mykola Lysenko <mykolal@...com>, linux-kernel@...r.kernel.org,
linux-kselftest@...r.kernel.org, kernel-team@...roid.com
Subject: Re: [PATCH 0/3] Dynptr Verifier Adjustments
> On Fri, Apr 28, 2023 at 2:04 AM Jose E. Marchesi
> <jose.marchesi@...cle.com> wrote:
>>
>>
>> > On Thu, Apr 27, 2023 at 4:36 PM Daniel Rosenberg <drosen@...gle.com> wrote:
>> >>
>> >> On Wed, Apr 26, 2023 at 4:39 PM Andrii Nakryiko
>> >> <andrii.nakryiko@...il.com> wrote:
>> >> >
>> >> > Don't know, show the sequence of commands you are running?
>> >> >
>> >> > I have linux source in ~/linux, and KBUILD_OUTPUT set to
>> >> > ~/linux-build/default. And it only takes this:
>> >> >
>> >> > $ cd ~/linux
>> >> > $ make -j90 # build kernel
>> >> > $ cd tools/testing/selftests/bpf
>> >> > $ make -j90 # build selftests
>> >> >
>> >> > And that's it.
>> >>
>> >> I've tried the same, modulo some paths. I'm pretty sure it's version
>> >> related at this point.
>> >> The current issue I'm seeing is "error: indirect call in function,
>> >> which are not supported by eBPF" when using GCC-BPF for
>> >> progs/bind4_prog.c
>> >
>> > I don't think GCC-BPF is able to compile selftests properly just yet.
>> > So I guess the problem is that you do have some version of gcc-bpf in
>> > the system and selftest's Makefile tries to build gcc variants of
>> > test_progs? That's bad (I don't have GCC-BPF locally, and everyone
>> > else apparently as well).
>> >
>> > So for now just `make BPF_GCC=` ? CC'ing Jose, we should probably
>> > agree on some criteria of "GCC-BPF is really capable of building
>> > selftests" and adjust Makefile to only attempt GCC BPF build in that
>> > case.
>>
>> Being able to run the selftests is our goal at the moment, but we are
>> not there yet, no.
>>
>> What about making the kernel build system to emit a visible warning
>> before it builds the GCC variants of the tests programs? Something like
>> "this is experimental and will likely fail".
>
> Given gcc-bpf can't build selftests right now, should we just disable
> it until there is a version on which gcc-bpf works? We can make it
> such that you can force it to build using gcc-bpf (make USE_GCC_BPF=1
> or something).
I think that makes sense.
>>
>> >>
>> >> Currently using clang 16.0.0 and gcc 12.2.0-14.
>> >> I did manage to get it to build by just commenting out TEST_GEN_PROGS
>> >> += test_progs-bpf_gcc
Powered by blists - more mailing lists