On Fri, Jul 22, 2005 at 02:13:12PM -0700, Chris Lattner wrote: > Anyone have any thoughts or opinions? This may have impact on a > number of things, including any pointer alignment analysis.
MEM_ALIGN has code to deal with this. There's no reason you can't apply that to TYPE_ALIGN. I suspect that this will pessimize some vectorizer output wrt pointers, but there's no real helping that: we'll actually be more correct. r~