Contradiction in mttp description

[SiFiveHolland]

In section 3, the description of mttp includes the sentence "In this case, the remaining fields (SDID, PPN) in mttp must be set to zeros, else generate a fault." This statement contradicts the register diagram just above that paragraph, which describes these fields as WARL.

[eckhard-delfs-qualcomm]

Maybe rephrase according to priv spec 10.1.11 (no fault generation):
To select MODE=Bare, software must write zero to the remaining fields (SDID, PPN) in mttp.
Attempting to select MODE=Bare with a nonzero pattern in the remaining fields has an UNSPECIFIED effect on the value that the remaining fields assume and an UNSPECIFIED effect on supervisor domain protection behavior.

[ram2532]

I agree with @eckhard-delfs-qualcomm

The mttp is similar in definition and behavior as to satp so I'd prefer similar write behavior when MODE=Bare. In particular, I don't think it should take an exception.

[rock-ifly]

I don't think it should take an exception when mttp.mode=Bare while ((mttp.sdid != 0) | (mttp.ppn !=0)). Since we can just bypass the mttCheck when mttp.mode-Bare, just neglect mttp.sdid and mttp.ppn.

On one hand, the WARL should not raise an exception as spec says. On the other hand, we can do the same thing as satp. The satp won't throw an exception when (satp.mode =Bare) & ((satp.asid != 0) | (satp.ppn != 0) ).

[rsahita]

addressed by PR #95