On Sun, Jul 27, 2025 at 4:32 PM Daniel Borkmann <daniel@xxxxxxxxxxxxx> wrote:
>
> Move them into bpf.h given we also need them in core code, add also
> for_each_cgroup_storage_type_cond() which we'll be using in a
> subsequent change.
>
> Signed-off-by: Daniel Borkmann <daniel@xxxxxxxxxxxxx>
> ---
> include/linux/bpf-cgroup.h | 3 ---
> include/linux/bpf.h | 20 ++++++++++++--------
> 2 files changed, 12 insertions(+), 11 deletions(-)
>
> diff --git a/include/linux/bpf-cgroup.h b/include/linux/bpf-cgroup.h
> index 082ccd8ad96b..d1f01d1168a1 100644
> --- a/include/linux/bpf-cgroup.h
> +++ b/include/linux/bpf-cgroup.h
> @@ -77,9 +77,6 @@ to_cgroup_bpf_attach_type(enum bpf_attach_type attach_type)
> extern struct static_key_false cgroup_bpf_enabled_key[MAX_CGROUP_BPF_ATTACH_TYPE];
> #define cgroup_bpf_enabled(atype) static_branch_unlikely(&cgroup_bpf_enabled_key[atype])
>
> -#define for_each_cgroup_storage_type(stype) \
> - for (stype = 0; stype < MAX_BPF_CGROUP_STORAGE_TYPE; stype++)
> -
> struct bpf_cgroup_storage_map;
>
> struct bpf_storage_buffer {
> diff --git a/include/linux/bpf.h b/include/linux/bpf.h
> index a87646cc5398..a41dff574327 100644
> --- a/include/linux/bpf.h
> +++ b/include/linux/bpf.h
> @@ -208,6 +208,18 @@ enum btf_field_type {
> BPF_RES_SPIN_LOCK = (1 << 12),
> };
>
> +enum bpf_cgroup_storage_type {
> + BPF_CGROUP_STORAGE_SHARED,
> + BPF_CGROUP_STORAGE_PERCPU,
> + __BPF_CGROUP_STORAGE_MAX
> +#define MAX_BPF_CGROUP_STORAGE_TYPE __BPF_CGROUP_STORAGE_MAX
> +};
> +
> +#define for_each_cgroup_storage_type(stype) \
> + for (stype = 0; stype < MAX_BPF_CGROUP_STORAGE_TYPE; stype++)
> +#define for_each_cgroup_storage_type_cond(stype, cond) \
> + for (stype = 0; (cond) && stype < MAX_BPF_CGROUP_STORAGE_TYPE; stype++)
> +
tbh I don't like this hidden condition.
It makes usage of this macro harder to follow.
Just open coded it in the callsite ?
Unrelated to the patch, but Dual #define for_each_cgroup_storage_type
for CONFIG_CGROUP_BPF vs not is ugly.
I don't understand why we added it.
I see no harm doing it once since enum bpf_cgroup_storage_type
is unconditional.
--
pw-bot: cr
