[PATCH v2 bpf-next 5/5] bpf: Make bpf_cgroup_read_xattr available to cgroup and struct_ops progs

[Song Liu]

> On Jun 20, 2025, at 11:18 AM, Alexei Starovoitov <alexei.starovoitov at gmail.com> wrote:
> 
> On Thu, Jun 19, 2025 at 3:02 PM Song Liu <song at kernel.org> wrote:
>> 
>> cgroup BPF programs and struct_ops BPF programs (such as sched_ext), need
>> bpf_cgroup_read_xattr. Make bpf_cgroup_read_xattr available to these prog
>> types.
> 
> ...
> 
>> +       ret = register_btf_kfunc_id_set(BPF_PROG_TYPE_LSM, &bpf_lsm_fs_kfunc_set);
>> +       ret = ret ?: register_btf_kfunc_id_set(BPF_PROG_TYPE_STRUCT_OPS, &bpf_fs_kfunc_set);
>> +       ret = ret ?: register_btf_kfunc_id_set(BPF_PROG_TYPE_CGROUP_SKB, &bpf_fs_kfunc_set);
>> +       ret = ret ?: register_btf_kfunc_id_set(BPF_PROG_TYPE_CGROUP_SOCK, &bpf_fs_kfunc_set);
>> +       ret = ret ?: register_btf_kfunc_id_set(BPF_PROG_TYPE_CGROUP_DEVICE, &bpf_fs_kfunc_set);
>> +       ret = ret ?: register_btf_kfunc_id_set(BPF_PROG_TYPE_CGROUP_SOCK_ADDR, &bpf_fs_kfunc_set);
>> +       ret = ret ?: register_btf_kfunc_id_set(BPF_PROG_TYPE_CGROUP_SYSCTL, &bpf_fs_kfunc_set);
>> +       return ret ?: register_btf_kfunc_id_set(BPF_PROG_TYPE_CGROUP_SOCKOPT, &bpf_fs_kfunc_set);
> 
> No need to artificially restrict it like this.
> bpf_cgroup_read_xattr() is generic enough and the verifier will enforce
> the safety due to KF_RCU.
> Just add it to common_btf_ids.

Makes sense. I will add it to common_btf_ids in v3. 

Thanks,
Song