[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <68c81196-5421-8731-52be-b9b620837eb5@gmail.com>
Date: Wed, 11 Jan 2023 12:02:12 -0300
From: Martin Rodriguez Reboredo <yakoyoku@...il.com>
To: Eric Curtin <ecurtin@...hat.com>
Cc: linux-kernel@...r.kernel.org,
Arnaldo Carvalho de Melo <acme@...nel.org>,
Neal Gompa <neal@...pa.dev>, bpf@...r.kernel.org,
rust-for-linux@...r.kernel.org,
Alexei Starovoitov <ast@...nel.org>,
Daniel Borkmann <daniel@...earbox.net>,
Andrii Nakryiko <andrii@...nel.org>,
Jiri Olsa <jolsa@...nel.org>, Yonghong Song <yhs@...com>,
Miguel Ojeda <miguel.ojeda.sandonis@...il.com>
Subject: Re: [PATCH v2] scripts: Exclude Rust CUs with pahole
On 1/8/23 16:19, Eric Curtin wrote:
> On Sun, 8 Jan 2023 at 15:18, Eric Curtin <ecurtin@...hat.com> wrote:
>>
>> On Sun, 8 Jan 2023 at 02:15, Martin Rodriguez Reboredo
>> <yakoyoku@...il.com> wrote:
>>>
>>> [ ... ]
>>
>> If it was me, I would do things more like v1 of the patch (instead
>> just checking pahole version), because this is the only flag set in
>> scripts/Makefile.modfinal, which is a little confusing and
>> inconsistent. It's ok to set --lang_exclude=rust in all cases, as long
>> as pahole_ver is recent enough.
>>
>> +if [ "${pahole_ver}" -ge "124" ]; then
>> + # see PAHOLE_HAS_LANG_EXCLUDE
>> + extra_paholeopt="${extra_paholeopt} --lang_exclude=rust"
>> +fi
>>
>> But I'm not too opinionated either on this so...
>>
>> Reviewed-by: Eric Curtin <ecurtin@...hat.com>
>>
>> can be reapplied. I'm gonna test this again to see if it works in a
>> Fedora Asahi rpm build.
>
> After testing I probably have to retract my Reviewed-by tag,
> apologies, bpf and all that did not work with this patch when I built
> in the fedora way, but, the good news is when I alter v1 of the patch
> to just check pahole version like so (instead of the is_enabled
> check):
>
> +if [ "${pahole_ver}" -ge "124" ]; then
> + # see PAHOLE_HAS_LANG_EXCLUDE
> + extra_paholeopt="${extra_paholeopt} --lang_exclude=rust"
> +fi
>
> it worked just fine, and that should satisfy the testbot and all the
> other ways we build too. Could we change to that @Martin Rodriguez
> Reboredo ?
>
>From my POV I don't like this way due to it being set regardless whether
or not you are building the kernel with Rust. Though, because it doesn't
affect non `CONFIG_RUST` builds, I _think_ it won't hurt if we use that
way for now. Gonna send v3.
>>
>>
>>> +
>>> vmlinux_link ${1}
>>>
>>> info "BTF" ${2}
>>> --
>>> 2.39.0
>>>
>
Powered by blists - more mailing lists