Re: [PATCH bpf] bpf: fix configuration-dependent BTF function references

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Thu, Dec 12, 2024 at 3:00 PM Thomas Weißschuh <linux@xxxxxxxxxxxxxx> wrote:
>
> These BTF functions are not available unconditionally,
> only reference them when they are available.
>
> Avoid the following build warnings:
>
>   BTF     .tmp_vmlinux1.btf.o
> btf_encoder__tag_kfunc: failed to find kfunc 'bpf_send_signal_task' in BTF
> btf_encoder__tag_kfuncs: failed to tag kfunc 'bpf_send_signal_task'
>   NM      .tmp_vmlinux1.syms
>   KSYMS   .tmp_vmlinux1.kallsyms.S
>   AS      .tmp_vmlinux1.kallsyms.o
>   LD      .tmp_vmlinux2
>   NM      .tmp_vmlinux2.syms
>   KSYMS   .tmp_vmlinux2.kallsyms.S
>   AS      .tmp_vmlinux2.kallsyms.o
>   LD      vmlinux
>   BTFIDS  vmlinux
> WARN: resolve_btfids: unresolved symbol prog_test_ref_kfunc
> WARN: resolve_btfids: unresolved symbol bpf_crypto_ctx
> WARN: resolve_btfids: unresolved symbol bpf_send_signal_task
> WARN: resolve_btfids: unresolved symbol bpf_modify_return_test_tp
> WARN: resolve_btfids: unresolved symbol bpf_dynptr_from_xdp
> WARN: resolve_btfids: unresolved symbol bpf_dynptr_from_skb
>
> Signed-off-by: Thomas Weißschuh <linux@xxxxxxxxxxxxxx>
> ---
>  kernel/bpf/helpers.c  | 4 ++++
>  kernel/bpf/verifier.c | 8 ++++++++
>  2 files changed, 12 insertions(+)
>
> diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c
> index 751c150f9e1cd7f56e6a2b68a7ebb4ae89a30d2d..5edf5436a7804816b7dcf1bbef2624d71a985f20 100644
> --- a/kernel/bpf/helpers.c
> +++ b/kernel/bpf/helpers.c
> @@ -3089,7 +3089,9 @@ BTF_ID_FLAGS(func, bpf_task_get_cgroup1, KF_ACQUIRE | KF_RCU | KF_RET_NULL)
>  BTF_ID_FLAGS(func, bpf_task_from_pid, KF_ACQUIRE | KF_RET_NULL)
>  BTF_ID_FLAGS(func, bpf_task_from_vpid, KF_ACQUIRE | KF_RET_NULL)
>  BTF_ID_FLAGS(func, bpf_throw)
> +#ifdef CONFIG_BPF_EVENTS
>  BTF_ID_FLAGS(func, bpf_send_signal_task, KF_TRUSTED_ARGS)
> +#endif
>  BTF_KFUNCS_END(generic_btf_ids)
>
>  static const struct btf_kfunc_id_set generic_kfunc_set = {
> @@ -3135,7 +3137,9 @@ BTF_ID_FLAGS(func, bpf_dynptr_is_null)
>  BTF_ID_FLAGS(func, bpf_dynptr_is_rdonly)
>  BTF_ID_FLAGS(func, bpf_dynptr_size)
>  BTF_ID_FLAGS(func, bpf_dynptr_clone)
> +#ifdef CONFIG_NET
>  BTF_ID_FLAGS(func, bpf_modify_return_test_tp)
> +#endif

It makes little sense to have bpf_prog_test_run_tracing() and
bpf_modify_return_test_tp() depend on CONFIG_NET... It's just
historically where BPF_PROG_TEST_RUN functionality was implemented,
but it seems like we need to move bpf_prog_test_run_tracing() and
other tracing-related testing stuff into kernel/trace/bpf_trace.c or
somewhere under kernel/bpf/ (core.c? helpers.c?)

>  BTF_ID_FLAGS(func, bpf_wq_init)
>  BTF_ID_FLAGS(func, bpf_wq_set_callback_impl)
>  BTF_ID_FLAGS(func, bpf_wq_start)
> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> index 5e541339b2f6d1870561033fd55cca7144db14bc..77bbf58418fee7533bce539c8e005d2342ee1a48 100644
> --- a/kernel/bpf/verifier.c
> +++ b/kernel/bpf/verifier.c
> @@ -5526,7 +5526,9 @@ static bool in_rcu_cs(struct bpf_verifier_env *env)
>
>  /* Once GCC supports btf_type_tag the following mechanism will be replaced with tag check */
>  BTF_SET_START(rcu_protected_types)
> +#ifdef CONFIG_NET
>  BTF_ID(struct, prog_test_ref_kfunc)
> +#endif
>  #ifdef CONFIG_CGROUPS
>  BTF_ID(struct, cgroup)
>  #endif
> @@ -5534,7 +5536,9 @@ BTF_ID(struct, cgroup)
>  BTF_ID(struct, bpf_cpumask)
>  #endif
>  BTF_ID(struct, task_struct)
> +#ifdef CONFIG_CRYPTO
>  BTF_ID(struct, bpf_crypto_ctx)
> +#endif
>  BTF_SET_END(rcu_protected_types)
>
>  static bool rcu_protected_object(const struct btf *btf, u32 btf_id)
> @@ -11529,8 +11533,10 @@ BTF_ID(func, bpf_rdonly_cast)
>  BTF_ID(func, bpf_rbtree_remove)
>  BTF_ID(func, bpf_rbtree_add_impl)
>  BTF_ID(func, bpf_rbtree_first)
> +#ifdef CONFIG_NET
>  BTF_ID(func, bpf_dynptr_from_skb)
>  BTF_ID(func, bpf_dynptr_from_xdp)
> +#endif
>  BTF_ID(func, bpf_dynptr_slice)
>  BTF_ID(func, bpf_dynptr_slice_rdwr)
>  BTF_ID(func, bpf_dynptr_clone)
> @@ -11558,8 +11564,10 @@ BTF_ID(func, bpf_rcu_read_unlock)
>  BTF_ID(func, bpf_rbtree_remove)
>  BTF_ID(func, bpf_rbtree_add_impl)
>  BTF_ID(func, bpf_rbtree_first)
> +#ifdef CONFIG_NET
>  BTF_ID(func, bpf_dynptr_from_skb)
>  BTF_ID(func, bpf_dynptr_from_xdp)
> +#endif
>  BTF_ID(func, bpf_dynptr_slice)
>  BTF_ID(func, bpf_dynptr_slice_rdwr)
>  BTF_ID(func, bpf_dynptr_clone)
>
> ---
> base-commit: 5d287a7de3c95b78946e71d17d15ec9c87fffe7f
> change-id: 20241212-bpf-cond-ids-9bfbc64dd77b
>
> Best regards,
> --
> Thomas Weißschuh <linux@xxxxxxxxxxxxxx>
>





[Index of Archives]     [Linux Samsung SoC]     [Linux Rockchip SoC]     [Linux Actions SoC]     [Linux for Synopsys ARC Processors]     [Linux NFS]     [Linux NILFS]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]


  Powered by Linux