[PATCH] D19664: [MemorySSA] Port to new pass manager

[Daniel Berlin via llvm-commits]

Yeah, i don't either.
At this point, why don't we just have it own the caching walker and see
what it breaks.

I know we almost certainly want walkers that memoryssa doesn't own (see bug
27740 - the best way i can think of to do this is to have a walker that GVN
uses and understands GVN's specific value equivalence knowledge )


On Tue, May 31, 2016 at 5:31 PM, George Burgess IV <
george.burgess.iv at gmail.com> wrote:

> george.burgess.iv added inline comments.
>
> ================
> Comment at: include/llvm/Transforms/Utils/MemorySSA.h:592
> @@ -595,3 +591,3 @@
>    // Memory SSA building info
> -  MemorySSAWalker *Walker;
> +  std::unique_ptr<MemorySSAWalker> Walker;
>    unsigned NextID;
> ----------------
> FWIW, I think Danny wasn't a massive fan of MSSA owning the caching
> walker. I don't fully recall why, though.
>
>
> http://reviews.llvm.org/D19664
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20160601/01a68bc9/attachment.html>