Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20230515121521.30569-1-lmb@isovalent.com> <a29c604e-5a68-eed2-b581-0ad4687fda10@linux.dev>
 <CAN+4W8hixyHYOwYRh-3WedS-a0KTQk8VQ4JxqM8y-DQY-yjsNA@mail.gmail.com> <a453c3d4-5615-f445-17a8-92a1dc4282e6@linux.dev>
In-Reply-To: <a453c3d4-5615-f445-17a8-92a1dc4282e6@linux.dev>
From:   Lorenz Bauer <lmb@isovalent.com>
Date:   Thu, 18 May 2023 09:42:28 +0100
Message-ID: <CAN+4W8iDy8w=aVErrSKA1OqJ7Onv3eszYOKHLr+zucfHRuHsVg@mail.gmail.com>
Subject: Re: [PATCH bpf-next] bpf: btf: restore resolve_mode when popping the
 resolve stack
To:     Martin KaFai Lau <martin.lau@linux.dev>
Cc:     bpf@vger.kernel.org, linux-kernel@vger.kernel.org,
        Alexei Starovoitov <ast@kernel.org>,
        Daniel Borkmann <daniel@iogearbox.net>,
        Andrii Nakryiko <andrii@kernel.org>,
        Song Liu <song@kernel.org>, Yonghong Song <yhs@fb.com>,
        John Fastabend <john.fastabend@gmail.com>,
        KP Singh <kpsingh@kernel.org>,
        Stanislav Fomichev <sdf@google.com>,
        Hao Luo <haoluo@google.com>, Jiri Olsa <jolsa@kernel.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Thu, May 18, 2023 at 2:42=E2=80=AFAM Martin KaFai Lau <martin.lau@linux.=
dev> wrote:
>
> On 5/17/23 2:01 AM, Lorenz Bauer wrote:
> > On Wed, May 17, 2023 at 7:26=E2=80=AFAM Martin KaFai Lau <martin.lau@li=
nux.dev> wrote:
>
> I can see your point to refactor it to make it work for all different BTF=
_KIND.
>
> Other than BTF_KIND_DATASEC, env->resolve_mode stays the same for all oth=
er
> kinds once it is decided. It is why resolve_mode is in the "env" instead =
of "v".
> My concern is this will hide some bugs (existing or future) that accident=
ally
> changed the resolve_mode in the middle. If there is another legit case th=
at
> could be found other than BTF_KIND_DATASEC, that will be a better time to=
 do
> this refactoring with a proper test case considering most bpf progs need =
btf to
> load nowadays and probably need to veristat test also. If it came to that=
, might
> as well consider moving resolve_mode from "env" to "v".
>
> btf_datasec_resolve() is acting as a very top level resolver like btf_res=
olve(),
> so it reset env->resolve_mode before resolving its var member like how
> btf_resolve() does. imo, together with env->resolve_mode stays the same f=
or
> others, it is more straight forward to reason. I understand that it is pe=
rsonal
> preference and could argue either way.

 Okay, let's drop it then :)