Re: [PATCH bpf-next v9 2/3] bpf: Fix s16 truncation for large bpf-to-bpf call offsets

[Quentin Monnet <qmo@kernel.org>]

2026-04-30 23:00 UTC+0800 ~ Yazhou Tang <tangyazhou@zju.edu.cn>
> Hi Kuohai,
> 
> Thanks for your review and the Acked-by tag.
> 
> On 4/30/26 11:29, Xu Kuohai wrote:
>> On 4/30/2026 1:19 AM, Yazhou Tang wrote:
> 
> [...]
> 
>>> diff --git a/kernel/bpf/fixups.c b/kernel/bpf/fixups.c
>>> index 300e4e251931..8947ef74f6a8 100644
>>> --- a/kernel/bpf/fixups.c
>>> +++ b/kernel/bpf/fixups.c
>>> @@ -1250,9 +1250,9 @@ static int jit_subprogs(struct bpf_verifier_env
>>> *env)
>>>           }
>>>           if (!bpf_pseudo_call(insn))
>>>               continue;
>>> -        insn->off = env->insn_aux_data[i].call_imm;
>>> -        subprog = bpf_find_subprog(env, i + insn->off + 1);
>>> -        insn->imm = subprog;
>>> +        insn->imm = env->insn_aux_data[i].call_imm;
>>> +        subprog = bpf_find_subprog(env, i + insn->imm + 1);
>>> +        insn->off = subprog;
>>>       }
>>>       prog->jited = 1;
>>> diff --git a/tools/bpf/bpftool/cfg.c b/tools/bpf/bpftool/cfg.c
>>> index e3785f9a697d..df43a0e0023f 100644
>>> --- a/tools/bpf/bpftool/cfg.c
>>> +++ b/tools/bpf/bpftool/cfg.c
>>> @@ -142,7 +142,7 @@ static bool cfg_partition_funcs(struct cfg *cfg,
>>> struct bpf_insn *cur,
>>>               continue;
>>>           if (cur->src_reg != BPF_PSEUDO_CALL)
>>>               continue;
>>> -        func = cfg_append_func(cfg, cur + cur->off + 1);
>>> +        func = cfg_append_func(cfg, cur + cur->imm + 1);
>>
>> Do we need to matain backward compatibility with the old bpftool here?
>> I suspect not, but I'm not entirely sure.
> 
> I agree. Since bpftool is maintained in-tree, it should be enough to just
> keep it in sync with the current kernel tree.

Hi, bpftool is also mirrored on GitHub at
https://github.com/libbpf/bpftool
and many people build it from there, so yes, it should be backward
compatible if possible at all.

Thanks,
Quentin