[llvm-commits] [PATCH, PowerPC] Fix PR13949

Adhemerval Zanella azanella at linux.vnet.ibm.com
Tue Oct 16 05:39:34 PDT 2012


On 10/16/2012 09:32 AM, Bill Schmidt wrote:
>
> On Tue, 2012-10-16 at 09:18 -0300, Adhemerval Zanella wrote:
>> On 10/16/2012 12:04 AM, Bill Schmidt wrote:
>>> Index: test/CodeGen/PowerPC/structsinregs.ll
>>> ===================================================================
>>> --- test/CodeGen/PowerPC/structsinregs.ll	(revision 165819)
>>> +++ test/CodeGen/PowerPC/structsinregs.ll	(working copy)
>>> @@ -1,5 +1,9 @@
>>>  ; RUN: llc -mcpu=pwr7 -O0 -disable-fp-elim < %s | FileCheck %s
>>> Index: test/CodeGen/PowerPC/structsinmem.ll
>>> ===================================================================
>>> --- test/CodeGen/PowerPC/structsinmem.ll	(revision 0)
>>> +++ test/CodeGen/PowerPC/structsinmem.ll	(revision 0)
>>> @@ -0,0 +1,227 @@
>>> +; RUN: llc -mcpu=pwr7 -O0 -disable-fp-elim < %s | FileCheck %s
>> Hey Bill, why the 'pwr7' on the tests? I believe they don't need this constraint.
> Without it, I've noticed we can get different default CPUs for different
> buildbots, which can change the ordering of the instructions.  This
> provides more certainty for the results.
>
> Thanks,
> Bill

What about 'ppc64'? From 'PPCGenSubtargetInfo.inc' currently it has the same itinerary for 'pwr7'
(G5ItinerariesModel).




More information about the llvm-commits mailing list