bpf: Support struct btf_struct_meta via KF_IMPLICIT_ARGS

[kernel-patches-daemon-bpf]

Pull request for series with
subject: bpf: Support struct btf_struct_meta via KF_IMPLICIT_ARGS
version: 1
url: https://patchwork.kernel.org/project/netdevbpf/list/?series=1065888

[kernel-patches-daemon-bpf]

Upstream branch: ca0f39a
series: https://patchwork.kernel.org/project/netdevbpf/list/?series=1065888
version: 1

[kernel-patches-review-bot]

commit a8c144a47f9df6b37f18ab77c2434c232b710e9f
Author: Ihor Solodrai <ihor.solodrai@linux.dev>

bpf: Support struct btf_struct_meta via KF_IMPLICIT_ARGS

Add new kfuncs with KF_IMPLICIT_ARGS that correspond to each _impl
kfunc accepting meta__ign, with typed struct btf_struct_meta *meta
parameters. The existing _impl kfuncs become thin wrappers for
backwards compatibility.

Link: https://lore.kernel.org/bpf/20260120222638.3976562-1-ihor.solodrai@linux.dev/

> diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c
> index cb6d242bd093..e5c8fb6596e7 100644
> --- a/kernel/bpf/helpers.c
> +++ b/kernel/bpf/helpers.c

[ ... ]

> +__bpf_kfunc void *bpf_percpu_obj_new(u64 local_type_id__k, struct btf_struct_meta *meta)
> +{
> +	u64 size = local_type_id__k;
> +
> +	/* The verifier has ensured that meta__ign must be NULL */
> +	return bpf_mem_alloc(&bpf_global_percpu_ma, size);
> +}

The parameter here is named "meta", not "meta__ign". Should this
comment say "meta" instead?

[ ... ]

> +__bpf_kfunc void bpf_percpu_obj_drop(void *p__alloc, struct btf_struct_meta *meta)
> +{
> +	/* The verifier has ensured that meta__ign must be NULL */
> +	bpf_mem_free_rcu(&bpf_global_percpu_ma, p__alloc);
> +}

Same here -- the comment references "meta__ign" but the parameter is
"meta".

> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> index 4fbacd2149cd..2e4f1e9b1d37 100644
> --- a/kernel/bpf/verifier.c
> +++ b/kernel/bpf/verifier.c

[ ... ]

> @@ -13575,12 +13658,12 @@ static int check_kfunc_args(struct bpf_verifier_env *env, struct bpf_kfunc_call_
>  		case KF_ARG_PTR_TO_ALLOC_BTF_ID:
>  			if (reg->type == (PTR_TO_BTF_ID | MEM_ALLOC)) {
> -				if (meta->func_id != special_kfunc_list[KF_bpf_obj_drop_impl]) {
> +				if (!is_bpf_obj_drop_kfunc(meta->func_id)) {
>  					verbose(env, "arg#%d expected for bpf_obj_drop_impl()\n", i);
>  					return -EINVAL;
>  				}
>  			} else if (reg->type == (PTR_TO_BTF_ID | MEM_ALLOC | MEM_PERCPU)) {
> -				if (meta->func_id != special_kfunc_list[KF_bpf_percpu_obj_drop_impl]) {
> +				if (!is_bpf_percpu_obj_drop_kfunc(meta->func_id)) {
>  					verbose(env, "arg#%d expected for bpf_percpu_obj_drop_impl()\n", i);
>  					return -EINVAL;
>  				}

The conditions now match both bpf_obj_drop and bpf_obj_drop_impl (and
their percpu variants), but the error messages still reference only
the _impl names. Would it make sense to update these to say
"bpf_obj_drop" and "bpf_percpu_obj_drop" respectively, so that a
developer calling the new variants sees the right function name in
the error?

---

AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md

In-Reply-To-Subject: bpf: Support struct btf_struct_meta via KF_IMPLICIT_ARGS
CI run summary: https://github.com/kernel-patches/bpf/actions/runs/23020502189

[kernel-patches-review-bot]

commit 1cf438425c438fadc801d3c0024a225f7bcdff92
Author: Ihor Solodrai <ihor.solodrai@linux.dev>

selftests/bpf: Update kfuncs using btf_struct_meta to new variants

Update selftests to use the new non-_impl kfuncs marked with
KF_IMPLICIT_ARGS by removing redundant declarations and macros from
bpf_experimental.h and updating relevant callsites.

> diff --git a/tools/testing/selftests/bpf/bpf_experimental.h b/tools/testing/selftests/bpf/bpf_experimental.h
> --- a/tools/testing/selftests/bpf/bpf_experimental.h
> +++ b/tools/testing/selftests/bpf/bpf_experimental.h
> @@ -8,156 +8,11 @@ #define __BPF_EXPERIMENTAL__
>  #define __contains(name, node) __attribute__((btf_decl_tag("contains:" #name ":" #node)))

[ ... ]

> -extern void *bpf_percpu_obj_new_impl(__u64 local_type_id, void *meta) __ksym;
> +/* Convenience macro to wrap over bpf_obj_new */
> +#define bpf_obj_new(type) ((type *)bpf_obj_new(bpf_core_type_id_local(type)))
>
>  /* Convenience macro to wrap over bpf_percpu_obj_new_impl */
> -#define bpf_percpu_obj_new(type) ((type __percpu_kptr *)bpf_percpu_obj_new_impl(bpf_core_type_id_local(type), NULL))
> +#define bpf_percpu_obj_new(type) ((type __percpu_kptr *)bpf_percpu_obj_new(bpf_core_type_id_local(type)))

The comment above bpf_percpu_obj_new still says "wrap over
bpf_percpu_obj_new_impl" but the macro now wraps bpf_percpu_obj_new.

> -extern void bpf_percpu_obj_drop_impl(void *kptr, void *meta) __ksym;

The extern for bpf_percpu_obj_drop_impl is removed here, but later
in the same file the bpf_percpu_obj_drop macro was not updated:

    /* Convenience macro to wrap over bpf_obj_drop_impl */
    #define bpf_percpu_obj_drop(kptr) bpf_percpu_obj_drop_impl(kptr, NULL)

Was bpf_percpu_obj_drop missed? Every other kfunc in this commit was
moved from the _impl variant to the new KF_IMPLICIT_ARGS variant
(bpf_obj_new, bpf_obj_drop, bpf_percpu_obj_new, bpf_refcount_acquire,
bpf_list_push_front, bpf_list_push_back, bpf_rbtree_add), but
bpf_percpu_obj_drop still goes through bpf_percpu_obj_drop_impl.

The macro is used in test_bpf_ma.c in batch_percpu_alloc() and
batch_percpu_free().

---

AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md

In-Reply-To-Subject: selftests/bpf: Update kfuncs using btf_struct_meta to new variants
CI run summary: https://github.com/kernel-patches/bpf/actions/runs/23020502189

[kernel-patches-daemon-bpf]

Forwarding comment 4049639337 via email
In-Reply-To: 20260312193546.192786-1-ihor.solodrai@linux.dev
Patch: https://patchwork.kernel.org/project/netdevbpf/patch/20260312193546.192786-1-ihor.solodrai@linux.dev/

[kernel-patches-daemon-bpf]

Forwarding comment 4049653995 via email
In-Reply-To: 20260312193546.192786-2-ihor.solodrai@linux.dev
Patch: https://patchwork.kernel.org/project/netdevbpf/patch/20260312193546.192786-2-ihor.solodrai@linux.dev/

[kernel-patches-daemon-bpf]

Upstream branch: bb41fce
series: https://patchwork.kernel.org/project/netdevbpf/list/?series=1065888
version: 1

[kernel-patches-daemon-bpf]

Upstream branch: bb41fce
series: https://patchwork.kernel.org/project/netdevbpf/list/?series=1065888
version: 1

[kernel-patches-daemon-bpf]

Upstream branch: 202e42e
series: https://patchwork.kernel.org/project/netdevbpf/list/?series=1065888
version: 1

[kernel-patches-daemon-bpf]

Upstream branch: 6c8e1a9
series: https://patchwork.kernel.org/project/netdevbpf/list/?series=1065888
version: 1

[kernel-patches-daemon-bpf]

Upstream branch: 2364959
series: https://patchwork.kernel.org/project/netdevbpf/list/?series=1065888
version: 1

[kernel-patches-daemon-bpf]

Upstream branch: 2364959
series: https://patchwork.kernel.org/project/netdevbpf/list/?series=1065888
version: 1

[kernel-patches-daemon-bpf]

Upstream branch: 2364959
series: https://patchwork.kernel.org/project/netdevbpf/list/?series=1065888
version: 1

[kernel-patches-daemon-bpf]

Upstream branch: 2364959
series: https://patchwork.kernel.org/project/netdevbpf/list/?series=1065888
version: 1

[kernel-patches-daemon-bpf]

Upstream branch: 2364959
series: https://patchwork.kernel.org/project/netdevbpf/list/?series=1065888
version: 1

[kernel-patches-daemon-bpf]

Upstream branch: 77378da
series: https://patchwork.kernel.org/project/netdevbpf/list/?series=1065888
version: 1

[kernel-patches-daemon-bpf]

Upstream branch: 77378da
series: https://patchwork.kernel.org/project/netdevbpf/list/?series=1065888
version: 1