Reg2Mem pass

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

Reg2Mem pass

Seung Jae Lee
Hi, guys.

I'm interested in using reg2mem pass but not sure about the concept of it.
Can you let me know a reference about it, if any?
Forgive my ignorance. -_-;

Thanks,
Seung J. Lee
_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: Reg2Mem pass

Reid Spencer-2
On Fri, 2007-08-10 at 00:51 -0500, Seung Jae Lee wrote:
> Hi, guys.
>
> I'm interested in using reg2mem pass but not sure about the concept of it.
> Can you let me know a reference about it, if any?
> Forgive my ignorance. -_-;

The reg2mem pass undoes SSA form. That is, it turns SSA registers into
alloca instructions with loads and stores.

>
> Thanks,
> Seung J. Lee
> _______________________________________________
> LLVM Developers mailing list
> [hidden email]         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev

_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: Reg2Mem pass

Andrew Lenharth
On 8/10/07, Reid Spencer <[hidden email]> wrote:
> On Fri, 2007-08-10 at 00:51 -0500, Seung Jae Lee wrote:
> > Hi, guys.
> >
> > I'm interested in using reg2mem pass but not sure about the concept of it.
> > Can you let me know a reference about it, if any?
> > Forgive my ignorance. -_-;
>
> The reg2mem pass undoes SSA form. That is, it turns SSA registers into
> alloca instructions with loads and stores.

With the exception that there still are some phis.  I don't remember
the issues leading up still having them, but they are much easier to
deal with I recall (I think it ensures that the phi is only taking
values defined in the immediate predessor, rather than an arbirary
dominator.

Andrew
_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev