Signed-off-by: YenHaoChen <39526191+YenHaoChen@users.noreply.github.com>

Co-authored-by: Tim Newsome <tim@sifive.com>
Signed-off-by: YenHaoChen <39526191+YenHaoChen@users.noreply.github.com>

chain() -> get_chain()
chain_bit -> chain

dmode() -> get_dmode()
trigger_t.dmode_bit -> mcontrol_t.dmode

store() -> get_store()
store_bit -> store

load() -> get_load()
load_bit -> bit

execute() -> get_execute(), execute_bit -> execute

structuralize trigger's tdata CSRs for modularization

Fix minor MMU bugs; clean up MMU some more

Fix tval reporting for CBOs; constrain cache-block sizes to reasonable values

mmu_t::translate may raise exceptions and so must be called with the
original virtual address for tval to be set correctly.

16 B suffices to subsume all aligned accesses (including the Q extension).
Spike does not actually rely on this property, but in some real systems,
it is impractical to guarantee atomicity across cache lines.

4096 B suffices to prevent cache lines from spanning pages (which would
require multiple TLB accesses).  This one is a bug fix, since we were
not performing multiple TLB accesses in this case.

Only triggers.cc uses the arithmetic function cto(). Instead of putting
the cto() in processor.h, putting it in arith.h with other arithmetic
functions, e.g., ctz() and clz(), makes more sense.

Invoking refill_tlb to get the physical address is somewhat baroque,
and not even helpful in practice, since successful SCs will fill
the TLB anyway.

The last remaining use was LR, which we can identify through other means.

The old implementation performed the load before checking whether the
memory region was valid for LR. So, for LR to MMIO, we would action
side effects before raising the exception, which is not precise.

For now, preserve the macro for compatibility with existing uses.

For now, preserve the macro for compatibility with existing uses.

The earlier call to store_slow_path will catch the misalignment.

These are now only used by AMOs.  Since AMOs are relatively uncommon,
and since the slow path really isn't slow anymore, send them down the
slow path to simplify the common case.