[llvm-commits] [llvm] r37843 - /llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp

[Evan Cheng]

On Jul 9, 2007, at 8:48 AM, Dan Gohman wrote:

> I found a problem in an interaction between my patch and the
> PPCISelLowering.cpp code for lowering call results.  The PPC target  
> code
> effectively assigns registers for a call result in reverse order. My
> patch was presenting it with registers already in target order.  
> This is
> more consistent with code for other constructs, so it seems best to  
> fix
> the PPCISelLowering.cpp code.

Ah, thanks for looking into this. I'd suspected something like this  
but didn't have time to dig in.

>
> Attached is a new patch. It's the same change to  
> SelectionDAGISel.cpp as
> before, but it's now accompanied by the PPCISelLowering.cpp update  
> and a
> test case which demonstrates the failure with the previous patch and
> passes with the new one.

I'll try it out asap.

>
> Thanks for your help,

You bet!

Evan

>
> Dan
>
> On Fri, Jul 06, 2007 at 01:40:33PM -0700, Evan Cheng wrote:
>> This seems to have fixed llvm-test failures, but Mac OS X PPC
>> bootstrapping is still unhappy. :-(
>>
>> Evan
>>
>
> -- 
> Dan Gohman, Cray Inc.
> <patch>