[llvm-commits] First patch to add support for variable address spaces to optimizers

[Villmow, Micah]

> -----Original Message-----
> From: Eli Friedman [mailto:eli.friedman at gmail.com]
> Sent: Wednesday, October 10, 2012 4:25 PM
> To: Villmow, Micah
> Cc: llvm-commits at cs.uiuc.edu
> Subject: Re: [llvm-commits] First patch to add support for variable
> address spaces to optimizers
> 
> On Wed, Oct 10, 2012 at 4:00 PM, Villmow, Micah <Micah.Villmow at amd.com>
> wrote:
> > Yeah, not sure which ones I need to modify though, then ones where I
> default to 0 I don't really see how I can get the address space, so I
> set it to the default.
> 
> I was thinking for the places where we're looking for the size of e.g.
> a pointer to a label (like in MachineFunction.cpp), you could add a
> special method "get the pointer size for the address space for the
> generated code is in".  Actually, I just realized that probably
> shouldn't be a global target setting anyway, and it isn't important for
> the first iteration, so ignore me. :)
[Villmow, Micah] Yeah I thought about this for awhile when I originally did this, but I can't really think of a case where a label would not be in the default address space. I mean I'm sure someone probably could do it, but it doesn't make sense to do so.
> 
> (The patch looks fine otherwise.)
> 
> -Eli