[llvm-commits] [PATCH] GVN patch for overflow intrinsics

[Chris Lattner]

On Jul 7, 2011, at 2:19 AM, Lang Hames wrote:

> Enhance GVN to look though extractvalues for recognized intrinsics. This is a first step towards fixing http://llvm.org/bugs/show_bug.cgi?id=8817 .
> 
> Could someone more familiar with GVN review this and let me know whether it's reasonable to commit?

Hi Lang,

I think that Owen is the best one to handle this.  In addition to handling the signed versions of these as well, does this correctly handle the case when the subtract comes first, and the cases when the overflow bit is actually used?

In principle, we'd want to optimize:

a = add i32 y, z
...
b,c = addo(y,z)

into:

b,c = addo(y,z)
a = b
...

Unfortunately, I'm not familiar enough with the GVN value table stuff to check the details of the patch, thanks for working on this though!

-Chris

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20110707/407ff059/attachment.html>