On Thu, 9 Jun 2022 at 23:58, Richard Henderson <richard.hender...@linaro.org> wrote: > > On 6/9/22 08:35, Peter Maydell wrote: > >> + if (!disas_sme(s, insn)) { > >> + unallocated_encoding(s); > >> + } > >> + break; > > > > I still think we should check bit 31 here. > > We don't do anything similar over in a32, where we've done the full > conversion: > > if (disas_a32_uncond(s, insn) || > > disas_vfp_uncond(s, insn) || > > disas_neon_dp(s, insn) || > > disas_neon_ls(s, insn) || > > disas_neon_shared(s, insn)) { > > return; > > }
The difference there is that we've basically completed the decodetree conversion. Here we're putting one decodetree disas call inside a legacy decode, so we should put it at the right point in the legacy structure, I think. > and there's nothing else within major opcode 0 to conflict. Yet :-) -- PMM