A use-after-free issue has been identified in the Linux kernel related to the bpf struct ops when CONFIG MODULES=n. The problem arises because the btf id of the "struct module" is missing, which prevents bpf try module get() from performing correct refcounting for struct ops that have a "struct module *owner" member. This issue affects struct ops that require the "struct module *owner" member, such as tcp congestion ops, but not all subsystems' struct ops, like sched ext ops. The issue has existed since the introduction of bpf struct ops and has undergone many changes.