Spec does not define "should"

[Timmmm]

Most formal specs I've read define words like "should", "must", "may", up front, but the RISC-V spec does not appear to. Despite that "should" appears dozens of times, often in places where it looks like they really meant "must", e.g.

In systems without S-mode, the medeleg and mideleg registers should not exist.

An Exception Code of 0 is reserved to mean "unknown cause" and implementations that do not distinguish sources of NMIs via the mcause register should return 0 in the Exception Code.

I think it would be helpful to audit those and probably change them to must.

Ideally adopt RFC2119 definitions and use them.

[ethindp]

It might be worth adopting RFC 2119 plus the RFC 8174 updates, requiring that those words appear in capitals to have the intended effect.