[all-commits] [llvm/llvm-project] 48ebe4: [SLPVectorizer] Make aliasing check more precise

[Nikita Popov via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 48ebe427c9c516c406f6fc1e4883bafaf5f4f992
      https://github.com/llvm/llvm-project/commit/48ebe427c9c516c406f6fc1e4883bafaf5f4f992
  Author: Nikita Popov <nikita.ppv at gmail.com>
  Date:   2021-08-31 (Tue, 31 Aug 2021)

  Changed paths:
    M llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
    M llvm/test/Transforms/SLPVectorizer/int_sideeffect.ll

  Log Message:
  -----------
  [SLPVectorizer] Make aliasing check more precise

SLPVectorizer currently uses AA::isNoAlias() to determine whether
two locations alias. This does not work if one of the instructions
is a call. Instead, we should check getModRefInfo(), which
determines whether an arbitrary instruction modifies or references
a given location.

Among other things, this prevents @llvm.experimental.noalias.scope.decl()
and other inaccessiblmemonly intrinsics from interfering with SLP
vectorization.

Differential Revision: https://reviews.llvm.org/D109012