[LLVMdev] [PATCH] CalculateSpillWeights does not need to be a pass

[Arnaud Allard de Grandmaison]

I had to revert it as it made 3 tests fail (ppc, arm). I will investigate
the failure when I will be back home --- I have to catch my plane now :).
Although there was no functional change intended, it changed register
allocation on those targets. Removing the CalculateSpillWeight pass from
the pipeline somehow affected the results of the pass pipeline on those
targets. May be those tests' checks just need to be fine tuned to be immune
to register allocation changes.

Cheers,
--
Arnaud


On Fri, Nov 8, 2013 at 7:05 PM, Arnaud Allard de Grandmaison <
arnaud.adegm at gmail.com> wrote:

> Committed at r194269
>
> Thanks Jakob !
>
> Cheers,
> --
> Arnaud
>
>
> On Fri, Nov 8, 2013 at 6:34 PM, Jakob Stoklund Olesen <stoklund at 2pi.dk>wrote:
>
>>
>> On Nov 8, 2013, at 7:40 AM, Arnaud Allard de Grandmaison <
>> arnaud.adegm at gmail.com> wrote:
>>
>> > Based on discussions with Lang Hames and Jakob Stoklund Olesen at the
>> hacker's lab, and in the light of upcoming work on the PBQP register
>> allocator, it was though that CalcSpillWeights does not need to be a pass.
>> This change will enable to customize / tune the spill weight computation
>> depending on the allocator.
>> >
>> > Update the documentation style while there.
>> >
>> > I also intend to take VirtRegAuxInfo out of line in a later patch as
>> this is really a helper class for spill weight (re) computation.
>> >
>> > Cheers,
>> > --
>> > Arnaud
>> > <0001-CalculateSpillWeights-does-not-need-to-be-a-pass.patch>
>>
>> LGTM.
>>
>> Thanks,
>> /jakob
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20131108/2d8de329/attachment.html>