[PATCH] D111003: [SCEV] Use full logic when infering flags on add and gep

[Philip Reames via Phabricator via llvm-commits]

reames created this revision.
reames added reviewers: efriedma, nikic.
Herald added subscribers: bollu, hiraditya, nemanjai, mcrosier.
Herald added a reviewer: bollu.
reames requested review of this revision.
Herald added a project: LLVM.

This is a followon to D109845 <https://reviews.llvm.org/D109845>.  Once that lands, 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.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D111003

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

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D111003.376721.patch
Type: text/x-patch
Size: 8486 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20211002/1d55596e/attachment.bin>