[PATCH] D20560: [new PM] port LoopAccessAnalysis to new pass manager (part-1)

[Xinliang David Li via llvm-commits]

On Tue, Jun 7, 2016 at 3:53 PM, Adam Nemet via llvm-commits <
llvm-commits at lists.llvm.org> wrote:

> anemet added a subscriber: anemet.
>
> ================
> Comment at: lib/Transforms/Scalar/LoopDistribute.cpp:611
> @@ -610,3 +610,3 @@
>      // LAA will check that we only have a single exiting block.
> -    LAI = &LAA->getInfo(L, ValueToValueMap());
> +    LAI = &LAA->getInfo().getInfo(L, ValueToValueMap());
>
> ----------------
> getInfo().getInfo(...), really?
>

yes -- that sounds kind of strange. I can make it similar to the new PM:

getResult().getInfo(..)

Do you have other preference?



>
> What is the difference between LAFI and LAA conceptually?
>


The difference is that LAA which is the legacy analysis pass will go away
in the future, but LAFI will stay as cached analysis results maintained by
new PM..



>
> Without knowing much about the new PM, I see that you split out the cache
> from LAA (which was pretty much all it contained).
>
> Should this perhaps be:
>
> LAA->getCache().getInfo(..)
>
> or getLoopInfo(..)
>
>
getLoopInfo is not a good name as it has other meaning.  I can change it to
getLoopAccessInfo, but I think getInfo should be fine if we change from
LAA->getInfo() to LAA->getResult().

David



> ?
>
>
> http://reviews.llvm.org/D20560
>
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20160607/ffb728f1/attachment.html>