[PATCH] D111257: [fir] Add affine demotion pass

[Valentin Clement via Phabricator via llvm-commits]

clementval created this revision.
clementval added reviewers: jeanPerier, svedanayagam, sscalpone, kiranchandramohan, jdoerfert, schweitz, pmccormick, mehdi_amini, rovka, AlexisPerry.
Herald added subscribers: Groverkss, mgorny.
Herald added a project: Flang.
clementval requested review of this revision.
Herald added a project: LLVM.
Herald added a subscriber: llvm-commits.

Add affine demotion pass.
Affine dialect's default lowering for loads and stores is different from
fir as it uses the `memref` type. The `memref` type is not compatible with
the Fortran runtime. Therefore, conversion of memory operations back to
`fir.load` and `fir.store` with `!fir.ref<?>` types is required.

This patch is part of the upstreaming effort from fir-dev branch.

Co-authored-by: Jean Perier <jperier at nvidia.com>
Co-authored-by: Eric Schweitz <eschweitz at nvidia.com>
Co-authored-by: V Donaldson <vdonaldson at nvidia.com>
Co-authored-by: Rajan Walia <walrajan at gmail.com>
Co-authored-by: Sourabh Singh Tomar <SourabhSingh.Tomar at amd.com>


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D111257

Files:
  flang/include/flang/Optimizer/Transforms/Passes.h
  flang/include/flang/Optimizer/Transforms/Passes.td
  flang/lib/Optimizer/Transforms/AffineDemotion.cpp
  flang/lib/Optimizer/Transforms/CMakeLists.txt
  flang/lib/Optimizer/Transforms/PassDetail.h
  flang/test/Fir/affine-demotion.fir

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D111257.377645.patch
Type: text/x-patch
Size: 13248 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20211006/d31f5325/attachment.bin>