Adding libbpf-rs maintainer, Daniel, for awareness, as Rust skeleton
will have to add support for this, once this patch set lands upstream.
On Mon, May 26, 2025 at 9:22 AM Leon Hwang <leon.hwang@xxxxxxxxx> wrote:
>
> This patch enhances bpftool to generate skeletons that properly handle
> global percpu variables. The generated skeleton now includes a dedicated
> structure for percpu data, allowing users to initialize and access percpu
> variables more efficiently.
>
> For global percpu variables, the skeleton now includes a nested
> structure, e.g.:
>
> struct test_global_percpu_data {
> struct bpf_object_skeleton *skeleton;
> struct bpf_object *obj;
> struct {
> struct bpf_map *data__percpu;
> } maps;
> // ...
> struct test_global_percpu_data__data__percpu {
> int data;
> char run;
> struct {
> char set;
> int i;
> int nums[7];
> } struct_data;
> int nums[7];
> } __aligned(8) *data__percpu;
>
> // ...
> };
>
> * The "struct test_global_percpu_data__data__percpu *data__percpu" points
> to initialized data, which is actually "maps.data__percpu->mmaped".
> * Before loading the skeleton, updating the
> "struct test_global_percpu_data__data__percpu *data__percpu" modifies
> the initial value of the corresponding global percpu variables.
> * After loading the skeleton, accessing or updating this struct is not
> allowed because this struct pointer has been reset as NULL. Instead,
> users must interact with the global percpu variables via the
> "maps.data__percpu" map.
>
> Signed-off-by: Leon Hwang <leon.hwang@xxxxxxxxx>
> ---
> tools/bpf/bpftool/gen.c | 47 +++++++++++++++++++++++++++++------------
> 1 file changed, 34 insertions(+), 13 deletions(-)
>
> diff --git a/tools/bpf/bpftool/gen.c b/tools/bpf/bpftool/gen.c
> index 67a60114368f5..c672f52110221 100644
> --- a/tools/bpf/bpftool/gen.c
> +++ b/tools/bpf/bpftool/gen.c
> @@ -92,7 +92,7 @@ static void get_header_guard(char *guard, const char *obj_name, const char *suff
>
> static bool get_map_ident(const struct bpf_map *map, char *buf, size_t buf_sz)
> {
> - static const char *sfxs[] = { ".data", ".rodata", ".bss", ".kconfig" };
> + static const char *sfxs[] = { ".data..percpu", ".data", ".rodata", ".bss", ".kconfig" };
> const char *name = bpf_map__name(map);
> int i, n;
>
> @@ -117,7 +117,7 @@ static bool get_map_ident(const struct bpf_map *map, char *buf, size_t buf_sz)
>
> static bool get_datasec_ident(const char *sec_name, char *buf, size_t buf_sz)
> {
> - static const char *pfxs[] = { ".data", ".rodata", ".bss", ".kconfig" };
> + static const char *pfxs[] = { ".data..percpu", ".data", ".rodata", ".bss", ".kconfig" };
> int i, n;
>
> /* recognize hard coded LLVM section name */
> @@ -148,7 +148,8 @@ static int codegen_datasec_def(struct bpf_object *obj,
> struct btf *btf,
> struct btf_dump *d,
> const struct btf_type *sec,
> - const char *obj_name)
> + const char *obj_name,
> + bool is_percpu)
> {
> const char *sec_name = btf__name_by_offset(btf, sec->name_off);
> const struct btf_var_secinfo *sec_var = btf_var_secinfos(sec);
> @@ -228,7 +229,7 @@ static int codegen_datasec_def(struct bpf_object *obj,
>
> off = sec_var->offset + sec_var->size;
> }
> - printf(" } *%s;\n", sec_ident);
> + printf(" }%s *%s;\n", is_percpu ? " __aligned(8)" : "", sec_ident);
> return 0;
> }
>
> @@ -263,13 +264,13 @@ static bool is_mmapable_map(const struct bpf_map *map, char *buf, size_t sz)
> return true;
> }
>
> - if (!bpf_map__is_internal(map) || !(bpf_map__map_flags(map) & BPF_F_MMAPABLE))
> - return false;
> -
> - if (!get_map_ident(map, buf, sz))
> - return false;
> + if (bpf_map__is_internal(map) &&
> + ((bpf_map__map_flags(map) & BPF_F_MMAPABLE) ||
> + bpf_map__is_internal_percpu(map)) &&
> + get_map_ident(map, buf, sz))
> + return true;
>
> - return true;
> + return false;
> }
>
> static int codegen_datasecs(struct bpf_object *obj, const char *obj_name)
> @@ -303,7 +304,8 @@ static int codegen_datasecs(struct bpf_object *obj, const char *obj_name)
> printf(" struct %s__%s {\n", obj_name, map_ident);
> printf(" } *%s;\n", map_ident);
> } else {
> - err = codegen_datasec_def(obj, btf, d, sec, obj_name);
> + err = codegen_datasec_def(obj, btf, d, sec, obj_name,
> + bpf_map__is_internal_percpu(map));
> if (err)
> goto out;
> }
> @@ -795,7 +797,8 @@ static int gen_trace(struct bpf_object *obj, const char *obj_name, const char *h
> bpf_object__for_each_map(map, obj) {
> const char *mmap_flags;
>
> - if (!is_mmapable_map(map, ident, sizeof(ident)))
> + if (!is_mmapable_map(map, ident, sizeof(ident)) ||
> + bpf_map__is_internal_percpu(map))
> continue;
>
> if (bpf_map__map_flags(map) & BPF_F_RDONLY_PROG)
> @@ -1434,7 +1437,25 @@ static int do_skeleton(int argc, char **argv)
> static inline int \n\
> %1$s__load(struct %1$s *obj) \n\
> { \n\
> - return bpf_object__load_skeleton(obj->skeleton); \n\
> + int err; \n\
> + \n\
> + err = bpf_object__load_skeleton(obj->skeleton); \n\
> + if (err) \n\
> + return err; \n\
> + \n\
> + ", obj_name);
> +
> + if (map_cnt) {
> + bpf_object__for_each_map(map, obj) {
> + if (bpf_map__is_internal_percpu(map) &&
> + get_map_ident(map, ident, sizeof(ident)))
> + printf("\tobj->%s = NULL;\n", ident);
> + }
> + }
hm... maybe we can avoid this by making libbpf re-mmap() this
initialization image to be read-only during bpf_object load? Then the
pointer can stay in the skeleton and be available for querying of
"initialization values" (if anyone cares), and we won't have any extra
post-processing steps in code generated skeleton code?
And Rust skeleton will be able to expose this as a non-mutable
reference with no extra magic behind it?
> +
> + codegen("\
> + \n\
> + return 0; \n\
> } \n\
> \n\
> static inline struct %1$s * \n\
> --
> 2.49.0
>
