[all-commits] [llvm/llvm-project] d02db3: [SCEV] Use full logic when infering flags on add a...

[Philip Reames via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: d02db32644b7360bcda54cdf739fa42abe450fcd
      https://github.com/llvm/llvm-project/commit/d02db32644b7360bcda54cdf739fa42abe450fcd
  Author: Philip Reames <listmail at philipreames.com>
  Date:   2021-10-03 (Sun, 03 Oct 2021)

  Changed paths:
    M llvm/lib/Analysis/ScalarEvolution.cpp
    M llvm/test/Analysis/Delinearization/multidim_ivs_and_integer_offsets_3d.ll
    M llvm/test/Analysis/Delinearization/multidim_ivs_and_parameteric_offsets_3d.ll
    M llvm/test/Analysis/LoopCacheAnalysis/PowerPC/stencil.ll
    M llvm/test/Analysis/ScalarEvolution/flags-from-poison.ll
    M llvm/test/Analysis/ScalarEvolution/load.ll
    M llvm/test/Analysis/ScalarEvolution/ptrtoint.ll
    M polly/test/IstAstInfo/simple-run-time-condition.ll

  Log Message:
  -----------
  [SCEV] Use full logic when infering flags on add and gep

This is a followon to D109845. With that landed, we will have fixed all known instances of pr51817, and can thus start inferring flags more aggressively with greatly reduced risk of miscompiles. This patch simply applies the same inference logic used in that patch to our other major flag inference path.

We can still do much better here (on both paths), but this is our first step.

Differential Revision: https://reviews.llvm.org/D111003