On 02/23/2018 06:18 AM, Peter Maydell wrote: >> + mm = (mm & 0xff) * (-1ull / 0xff); > > What is this expression doing? I guess from context that it's > replicating the low 8 bits of mm across the 64-bit value, > but this is too obscure to do without a comment or wrapping > it in a helper function with a useful name, I think.
I do have a helper now -- dup_const. I thought I'd converted all of the uses, but clearly missed one/some. r~