and data fixups performing shift/high adjust/sign extension on fieldval. Sink fx_pcrel handling and checks. Use fixP->fx_size when writing data fixups rather than recalculating size.