Re: [AD] AMD64 patch

[Peter Wang <tjaden@xxxxxxxxxx>]

Eric Botcazou wrote:

> > In other places where it didn't really matter either way, I thought it
> > would be best just to leave them alone.  Then the assumption on the
> > sizeof(int) can remain as it is now (somewhere between 16-bits and
> > 32-bits) and long can continue to be used to hold values which need
> > 32-bits.  Is there any reason to prefer a 32-bit integer over a 64-bit
> > integer on a 64-bit machine, e.g. to hold single pixel values?
> >    
>
> Alignment I'd say.  Every other 32-bit pixel value is not 64-bit aligned so 
> you must be careful when you're dereferencing a pointer to a scanline, 
> although this doesn't really matter on AMD64.
>  

Hmm, then I guess we should use ints internally to hold single pixel 
values but retain unsigned longs wherever needed in the interface for 
API compatibility?  That can be done after the current patch is applied.

If you could explain your thinking about the alignment a bit more, that 
would be great.  To me it seems like there is not much difference 
between dereferencing a 32-bit quantity and (i) storing it into a 32-bit 
register, or (ii) storing it into a 64-bit register.  Surely the only 
difference is that in the latter case the upper 32-bits of the register 
need to be cleared (or sign-extended)?

Peter