mdbgen as global input instead of per-hart

[gokhankaplayan]

An input port, named mdbgen[i], is introduced for each hart i in Chapter 2 (ISA-part). I propose to have global input port mdben instead of per-hart mdbgen.
There is no need for per-hart control in homogeneous computing.

[vicky-goode-img]

If this proposal is rejected and mdbgen remains per hart, then in the Debug Module (non_ISA) chapter the use of mdbgen needs to be clarified, e.g

• for hartreset and halt it should use mdbgen[i] instead of mdbgen
• for ndmreset it should probably say "Issuing ndmreset operation results in a security fault error when any bit of mdbgen is 0"

And possibly other cases in that chapter

[AoteJin]

I think we should take heterogeneous scenarios into consideration as well. A certain SoC might have many varied profiles of RISC-V hart serving different purposes. And each of them can be of different security level, thus have various debug policy for M-mode. Do you think it makes sense?

[gokhankaplayan]

Since the RISC-V Spec is generic, It makes sense to keep per-hart mdbgen[i] input in the Spec. Homogenous scenarios can implement global mdbgen input as an implementation choice.
As @vicky-goode-img mentioned we should be consistent to use mdben[i] in the document.