> Well, are you sure this is correct? I have been thinking they are different > for big endian. I've thought about it again, and think you're quite right. Another case where "rev" would be needed by the looks of it (if we derive our in-register representation from "ldr" this time). Cheers. Tim.