On 3/19/19 4:20 AM, Peter Maydell wrote: > Speaking of VECSTRIDE, in theory we should implement the v8A > requirement that FPSCR.Len and .Stride are either RAZ/WI or > cause the insns which use Len and Stride to UNDEF if they are > set to non-zero. At the moment we just have the v7A behaviour > of "honour Len and Stride settings"...
I was just thinking about those this morning. I was thinking of pushing the implementation of these completely to helper functions, so that we do not need to know the exact values during translate. We'd only need a single bit to trigger the use of the helpers (or a v8 undef). r~
