[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAEf4BzZk4onktrnK-i7CQUrFAPEo24G9p5RZhpg0nrhYxU5EvA@mail.gmail.com>
Date: Fri, 13 Sep 2024 13:46:31 -0700
From: Andrii Nakryiko <andrii.nakryiko@...il.com>
To: Tao Chen <chen.dylane@...il.com>
Cc: Andrii Nakryiko <andrii@...nel.org>, Eduard Zingerman <eddyz87@...il.com>,
Daniel Borkmann <daniel@...earbox.net>, Martin KaFai Lau <martin.lau@...ux.dev>, Song Liu <song@...nel.org>,
Yonghong Song <yonghong.song@...ux.dev>, John Fastabend <john.fastabend@...il.com>,
KP Singh <kpsingh@...nel.org>, Stanislav Fomichev <sdf@...ichev.me>, Hao Luo <haoluo@...gle.com>,
Jiri Olsa <jolsa@...nel.org>, bpf@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH bpf-next RESEND v2] libbpf: Fix expected_attach_type set
when kernel not support
On Fri, Sep 13, 2024 at 9:44 AM Tao Chen <chen.dylane@...il.com> wrote:
>
> The commit "5902da6d8a52" set expected_attach_type again with
> filed of bpf_program after libpf_prepare_prog_load, which makes
> expected_attach_type = 0 no sense when kenrel not support the
> attach_type feature, so fix it.
>
> Fixes: 5902da6d8a52 ("libbpf: Add uprobe multi link support to bpf_program__attach_usdt")
> Suggested-by: Jiri Olsa <jolsa@...nel.org>
> Signed-off-by: Tao Chen <chen.dylane@...il.com>
> ---
> tools/lib/bpf/libbpf.c | 6 ++----
> 1 file changed, 2 insertions(+), 4 deletions(-)
>
> Change list:
> - v1 -> v2:
> - restore the original initialization way suggested by Jiri
>
> diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
> index 219facd0e66e..df2244397ba1 100644
> --- a/tools/lib/bpf/libbpf.c
> +++ b/tools/lib/bpf/libbpf.c
> @@ -7353,7 +7353,7 @@ static int libbpf_prepare_prog_load(struct bpf_program *prog,
>
> /* special check for usdt to use uprobe_multi link */
> if ((def & SEC_USDT) && kernel_supports(prog->obj, FEAT_UPROBE_MULTI_LINK))
> - prog->expected_attach_type = BPF_TRACE_UPROBE_MULTI;
> + opts->expected_attach_type = BPF_TRACE_UPROBE_MULTI;
>
Ok, took me a bit to understand what the issue is. But the above is
not quite correct, for the above case of setting
BPF_TRACE_UPROBE_MULTI we do want to record BPF_TRACE_UPROBE_MULTI in
prog->expected_attach_type, because user might want to query that
later.
So I agree with the part of the fix below, but here I think we need
*both* update opts' and prog's expected_attach_type, so we will have:
prog->expected_attach_type = BPF_TRACE_UPROBE_MULTI;
opts->expected_attach_type = BPF_TRACE_UPROBE_MULTI;
pw-bot: cr
> if ((def & SEC_ATTACH_BTF) && !prog->attach_btf_id) {
> int btf_obj_fd = 0, btf_type_id = 0, err;
> @@ -7443,6 +7443,7 @@ static int bpf_object_load_prog(struct bpf_object *obj, struct bpf_program *prog
> load_attr.attach_btf_id = prog->attach_btf_id;
> load_attr.kern_version = kern_version;
> load_attr.prog_ifindex = prog->prog_ifindex;
> + load_attr.expected_attach_type = prog->expected_attach_type;
>
> /* specify func_info/line_info only if kernel supports them */
> if (obj->btf && btf__fd(obj->btf) >= 0 && kernel_supports(obj, FEAT_BTF_FUNC)) {
> @@ -7474,9 +7475,6 @@ static int bpf_object_load_prog(struct bpf_object *obj, struct bpf_program *prog
> insns_cnt = prog->insns_cnt;
> }
>
> - /* allow prog_prepare_load_fn to change expected_attach_type */
> - load_attr.expected_attach_type = prog->expected_attach_type;
> -
> if (obj->gen_loader) {
> bpf_gen__prog_load(obj->gen_loader, prog->type, prog->name,
> license, insns, insns_cnt, &load_attr,
> --
> 2.25.1
>
Powered by blists - more mailing lists