target/riscv: Mismatch napot when mcontrol.maskmax is zero

[sunnyzhu-learning]

Reference release specification (https://github.com/riscv/riscv-debug-spec/blob/release/riscv-debug-release.pdf), in section 5.2.9 introduce maskmx：
A value of 0 indicates that only exact value matches are supported (one byte range).

so. when tdata1.type=2 and napot is unsupported, try_setup_single_match_trigger should skip match=1 .

[en-sc]

AFAIU, the approach suggested in the commit will not be suitable for some HW configurations (please, see the comment on riscv_enumerate_triggers()).

I'd suggest to look at set_trigger() instead.
In particular, tdata1_config_denied can reflect that the value in mcontrol.maskmax fits the requested range.
e.g.:

1. Rename trigger_request_info::tdata1_ignore_mask -> trigger_request_info::maskmax.
2. Add the something like the following to set_trigger():

const uint32_t type = get_field(tdata1, CSR_TDATA1_TYPE(riscv_xlen(target)));
if (type == CSR_TDATA1_TYPE_MCONTROL &&
		(get_field(tdata1, CSR_MCONTROL_MATCH) == CSR_MCONTROL_MATCH_NAPOT))
		tdata1_config_denied |= get_field(tdata1_rb, CSR_MCONTROL_MASKMAX) < (get_least_significant_zero_index(tdata2) + 1).
}

(get_least_significant_zero_index() is not yet implemented).

[en-sc]

What if the trigger supports multiple types and is disabled at the moment of enumeration?
AFAIU, you it will not be possible to use such trigger for a NAPOT match.

[sunnyzhu-learning]

thanks a lot, I will create new PR addressed this question