On Jun 17, 2005, at 4:29 PM, der Mouse wrote:

> Acquiring alignment restrictions, that is *definitely* a bug, to my
> mind.  memcpy() as specified does not have alignment restrictions, so
> an inlining of it that does is an invalid transformation - unless the
> compiler can prove those restrictions are always satisfied for that
> particular call.

You have misunderstood what happens in the memcpy() case.... it  
inlines when it knows (according to type rules) that the src and dst  
are aligned in a certain way, that the (constant) size is within some  
threshold, and appropriate instructions are available for performing  
the block move in a way that would be cheaper than a function call.

-- thorpej