Assert misa.B bit through --isa=...B...

[YenHaoChen]

This PR provides misa.B=1 by --isa=...B...

Reference: #1559

[aswaterman]

Seems like we are becoming inconsistent about how we handle these different cases, in multiple ways:

• For P, we use the switch statement to populate the Zp extensions, rather than the if statements at the bottom. (This is only a stylistic concern; we don't need to fix it right now, but I thought it was worth mentioning.)
• For A, we make it so that A implies Zaamo and Zalrsc, but we don't make Zaamo + Zalrsc imply A. This is inconsistent with what you propose we do for B. We should either delete the logic that makes Zba + Zbb + Zbs imply B, or add the logic so that Zaamo + Zalrsc imply A. I think people would prefer the latter option, so that misa will contain A if Zaamo+Zalrsc is specified.

[YenHaoChen]

Thank you. I was not aware of the inconsistency. I targeted the missing character 'b' in the variable 'all_subsets,' which supports --isa=...B...

Let's delete the logic that makes Zba + Zbb + Zbs imply B in this PR. I will provide another PR that discusses the implications of A and B.

[aswaterman]

Does this version still make B imply Zba+Zbb+Zbs? It isn’t clear to me how it works.

[YenHaoChen]

Does this version still make B imply Zba+Zbb+Zbs? It isn’t clear to me how it works.

Yes, we have B implying Zba+Zbb+Zbs (disasm/isa_parser.cc:333). The previous PR (#1559) provided the code for implication, which aims at the B extension. Unfortunately, it does not offer --isa=...B... functionality from the optional flag. This PR catches up with the missing.

[aswaterman]

Ah, got it.