[llvm] r291968 - NewGVN: Move leaders around properly to ensure we have a canonical dominating leader. Fixes PR 31613.

[Hans Wennborg via llvm-commits]

OK, merged together with r291979 in r292307.

Thanks,
Hans

On Fri, Jan 13, 2017 at 5:22 PM, Davide Italiano <davide at freebsd.org> wrote:
> On Fri, Jan 13, 2017 at 2:40 PM, Daniel Berlin via llvm-commits
> <llvm-commits at lists.llvm.org> wrote:
>> Author: dannyb
>> Date: Fri Jan 13 16:40:01 2017
>> New Revision: 291968
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=291968&view=rev
>> Log:
>> NewGVN: Move leaders around properly to ensure we have a canonical dominating leader. Fixes PR 31613.
>>
>> Summary:
>> This is a testcase where phi node cycling happens, and because we do
>> not order the leaders by domination or anything similar, the leader
>> keeps changing.
>>
>> Using std::set for the members is too expensive, and we actually don't
>> need them sorted all the time, only at leader changes.
>>
>> We could keep both a set and a vector, and keep them mostly sorted and
>> resort as necessary, or use a set and a fibheap, but all of this seems
>> premature.
>>
>> After running some statistics, we are able to avoid the vast majority
>> of sorting by keeping a "next leader" field.  Most congruence classes only have
>> leader changes once or twice during GVN.
>>
>> Reviewers: davide
>>
>
> I think this can be merged to stable, as it fixes the last known issue
> I was able to find on the codebases I tried (of course, not the last
> issue in general). I have the hope of making this available for
> broader testing internally (my customers) and given we branch on top
> of every x.0 release, that would save me a bit of pain keeping local
> patches in our downstream tree. Also it's good for people who want to
> try this but don't keep track of trunk (just trying to elaborate a bit
> in case I ask to merge other NewGVN related patches, hope it makes
> sense).