On Tue, Dec 14, 2021 at 6:31 PM Jakub Kicinski <kuba@xxxxxxxxxx> wrote: > > Now that the stage has been set and actors are in place > remove the header dependency between cgroup and bpf.h. > > This reduces the incremental build size of x86 allmodconfig > after bpf.h was touched from ~17k objects rebuilt to ~5k objects. > bpf.h is 2.2kLoC and is modified relatively often. > > Signed-off-by: Jakub Kicinski <kuba@xxxxxxxxxx> > --- > include/linux/bpf-cgroup.h | 4 +++- > 1 file changed, 3 insertions(+), 1 deletion(-) > > diff --git a/include/linux/bpf-cgroup.h b/include/linux/bpf-cgroup.h > index 12474516e0be..4c932d47e7f2 100644 > --- a/include/linux/bpf-cgroup.h > +++ b/include/linux/bpf-cgroup.h > @@ -2,7 +2,8 @@ > #ifndef _BPF_CGROUP_H > #define _BPF_CGROUP_H > > -#include <linux/bpf.h> > +#include <linux/bpf-cgroup-types.h> > +#include <linux/bpf-link.h> Borked rebase with stale header I guess ? Could you try just patch 1 with bpf-cgroup-types.h and do s/bpf.h/bpf-cgroup-types.h/ here in bpf-cgroup.h as patch 2 while moving cgroup_storage_type() function to bpf.h ? Patch 1 will add +#include <linux/bpf-cgroup-types.h> to linux/bpf.h, so cgroup_storage_type() will have everything it needs there. I could be still missing something.
