[all-commits] [llvm/llvm-project] 82b7a2: [flang] hlfir.assign default implementation

[jeanPerier via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 82b7a238cf11f3a4f9ab91ff1df157dbf2b0bae0
      https://github.com/llvm/llvm-project/commit/82b7a238cf11f3a4f9ab91ff1df157dbf2b0bae0
  Author: Jean Perier <jperier at nvidia.com>
  Date:   2022-12-07 (Wed, 07 Dec 2022)

  Changed paths:
    M flang/include/flang/Optimizer/HLFIR/HLFIRDialect.h
    M flang/include/flang/Optimizer/HLFIR/Passes.td
    M flang/lib/Optimizer/Builder/HLFIRTools.cpp
    M flang/lib/Optimizer/HLFIR/Transforms/ConvertToFIR.cpp
    A flang/test/HLFIR/assign-codegen.fir

  Log Message:
  -----------
  [flang] hlfir.assign default implementation

Add the default unoptimized implementation implementation
of hlfir.assign. It relies on the runtime for array assignment
and always makes a temp of the right hand side for arrays.

Assignment optimization will be done when all HLFIR pieces are in place
and aliasing analysis is available.

Differential Revision: https://reviews.llvm.org/D139426