Changes
- [A64] Leave the scalar NaN rule to the hardware where it already agrees (
089473c) - [A64] Keep the scalar FP NaN fixups out of line, and shorten them (
5f36967) - [A64] Gate the vector FMA NaN fixup on the result (
497bec2) - [A64] Gate the scalar NaN fixups on the result (
d88e3ea) - [A64] Skip a same-register truncation that only compares read (
882df33) - [A64] Use REV32 and ZIP for byte-order permutes and merges (
968f1ef) - [A64] Select into an aliased operand with BIT/BIF (
ef4b3a4) - [A64] Remap on the effective address and shorten guest addressing (
004ece7)
Full changelog: 10c8ae7...089473c