On Mon, Feb 17, 2025 at 07:14:42PM -0800, JP Kobryn wrote: > The processing of bpf cgroup stats is tied to the rstat actions of other > subsystems. Make changes to have them updated/flushed independently. > Give the cgroup_bpf struct its own cgroup_rstat instance and define a > new cgroup_rstat_ops instance specifically for the cgroup_bpf. Then > replace the kfunc status of the existing updated/flush api calls with > non-kfunc status. As an alternative, create new updated/flush kfuncs > specifically for bpf cgroups. In these new kfuncs, make use of the > bpf-specific rstat ops to plumb back in to the existing rstat routines. > Where applicable, use pre-processor conditionals to define bpf rstat > related stuff. > > Signed-off-by: JP Kobryn <inwardvessel@xxxxxxxxx> Since we deicided to keep bpf with the base stats, this patch will go away.
