[llvm-commits] [PATCH] Additional SD-Node Cleanup after Combining Pass

[Evan Cheng]

Sorry to nitpick. But I think it's better to follow the conventions of legalizer passes by putting the call to RemoveDeadNodes() at the end of DAGCombiner::Run().

Evan

On Apr 13, 2012, at 11:03 AM, Hal Finkel wrote:

> I've attached an updated patch, this time placing the call to remove
> dead nodes inside Combine(). Please review.
> 
> Thanks again,
> Hal
> 
> On Thu, 12 Apr 2012 22:24:37 -0700
> Evan Cheng <evan.cheng at apple.com> wrote:
> 
>> 
>> 
>> On Apr 8, 2012, at 1:08 PM, Hal Finkel <hfinkel at anl.gov> wrote:
>> 
>>> On Sun, 08 Apr 2012 19:57:15 +0200
>>> Duncan Sands <baldrick at free.fr> wrote:
>>> 
>>>> Hi Hal,
>>>> 
>>>>> Ping. (This is a one-line change, and I'd like to get this fixed
>>>>> for Gabor).
>>>> 
>>>> isn't it a bug that the DAG combiner isn't deleting dead nodes?
>>> 
>>> It is certainly a bug that *something* is not deleting the dead
>>> nodes. Are you suggesting that the call to
>>> RemoveDeadNodes() should actually come from Combine()? or that
>>> there is some deeper structural problem?
>>> 
>> 
>> Yes, it seems like the individual passes should be responsible for
>> deleting dead nodes. 
>> 
>> Evan
>> 
>>> Thanks again,
>>> Hal
>>> 
>>>> 
>>>> Ciao, Duncan.
>>>> _______________________________________________
>>>> llvm-commits mailing list
>>>> llvm-commits at cs.uiuc.edu
>>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>> 
>>> 
>>> 
>>> -- 
>>> Hal Finkel
>>> Postdoctoral Appointee
>>> Leadership Computing Facility
>>> Argonne National Laboratory
>>> _______________________________________________
>>> llvm-commits mailing list
>>> llvm-commits at cs.uiuc.edu
>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
> 
> 
> 
> -- 
> Hal Finkel
> Postdoctoral Appointee
> Leadership Computing Facility
> Argonne National Laboratory
> <sd_cleanup_2.patch>