For "depends on DEBUG_INFO_BTF" this does not happen: Kconfig turns an m upper bound into y for a bool symbol (scripts/kconfig/symbol.c, sym_calc_visibility()). I checked it with olddefconfig: with SCHED_CLASS_EXT, NET_SCH_BPF, PROBE_EVENTS_BTF_ARGS, BINFMT_MISC_BPF, IO_URING_BPF_OPS and DEBUG_INFO_BTF_MODULES enabled, switching DEBUG_INFO_BTF from y to m changes nothing but BPF_PRELOAD, which the series excludes on purpose.
Looking at the other uses turned up the reverse case, though: RUST and GENDWARFKSYMS say "depends on !DEBUG_INFO_BTF", and !m is m, which a bool again takes as y, so with =m those pahole restrictions no longer applied. v3 writes them as DEBUG_INFO_BTF=n (patch 8/9): https://lore.kernel.org/bpf/[email protected]/ Jay
