[PATCH] D38586: SLPVectorizer.cpp: Ensure SLPVectorizer can visit each block by dominated order

[NAKAMURA Takumi via Phabricator via llvm-commits]

chapuni updated this revision to Diff 118120.
chapuni retitled this revision from "SLPVectorizer.cpp: Avoid std::stable_sort() due to asymmetric comparator" to "SLPVectorizer.cpp: Ensure SLPVectorizer can visit each block by dominated order".
chapuni added a comment.

The test added. This comes from EmitAtomicUpdateValue() in CGAtomic.cpp.
I confirmed it fails on both libstdc++ and libc++ to me.

I will attempt to get rid of O(N^2) there, but I think it'd not be big deal unless preds exceeded hundreds.


Repository:
  rL LLVM

https://reviews.llvm.org/D38586

Files:
  llvm/trunk/lib/Transforms/Vectorize/SLPVectorizer.cpp
  llvm/trunk/test/Transforms/SLPVectorizer/X86/visit-dominated.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D38586.118120.patch
Type: text/x-patch
Size: 10614 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20171007/7eeff5c2/attachment.bin>