Most CPUs provide custom system-level cache maintenance instructions, such as flushing or invalidating the entire cache, which are typically used during boot-up or power-down sequences.
Has the RISC-V CHERI project considered introducing capability-aware versions of these system-level cache maintenance instructions?
Specifically, is it possible for such instructions to reference a capability register and perform permission checks on each cache line as they operate?
If not, could the lack of capability checks in system-level cache maintenance operations potentially introduce security vulnerabilities?