[llvm-commits] CVS: llvm/lib/Analysis/LoadValueNumbering.cpp

Chris Lattner sabre at nondot.org
Wed Jun 22 20:36:11 PDT 2005


On Mon, 20 Jun 2005, Andrew Lenharth wrote:

> Log message:
>
> prevent GCSE from forwarding stores to loads around vaarg.  This is uggly, and I am trying to fix the AliasInfo, as it should catch the problem instead.

This and other related changes (e.g. LICM.cpp) are wrong: these passes 
don't need to know about vaarg specifically..  Please overload the 
"mayWriteToMemory" method in the vaarg instruction class, and revert these 
related patches.

Thanks,

-Chris

>
> LoadValueNumbering.cpp |    4 ++++
> 1 files changed, 4 insertions(+)
>
>
> Index: llvm/lib/Analysis/LoadValueNumbering.cpp
> diff -u llvm/lib/Analysis/LoadValueNumbering.cpp:1.31 llvm/lib/Analysis/LoadValueNumbering.cpp:1.32
> --- llvm/lib/Analysis/LoadValueNumbering.cpp:1.31	Thu Apr 21 16:04:58 2005
> +++ llvm/lib/Analysis/LoadValueNumbering.cpp	Mon Jun 20 10:02:05 2005
> @@ -328,6 +328,10 @@
>       LoadInvalidatedInBBBefore = true;
>       break;
>     }
> +    if (dyn_cast<VAArgInst>(I)) {
> +      LoadInvalidatedInBBBefore = true;
> +      break;
> +    }
>   }
>
>   // Figure out if the load is invalidated between the load and the exit of the
>
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://mail.cs.uiuc.edu/mailman/listinfo/llvm-commits
>

-Chris

-- 
http://nondot.org/sabre/
http://llvm.cs.uiuc.edu/




More information about the llvm-commits mailing list