[all-commits] [llvm/llvm-project] a0150c: [flang] Canonicalize fir.array_coor for contiguous...

[Slava Zakharin via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: a0150ce371d1d808e0753c954618388fbe7ef7b7
      https://github.com/llvm/llvm-project/commit/a0150ce371d1d808e0753c954618388fbe7ef7b7
  Author: Slava Zakharin <szakharin at nvidia.com>
  Date:   2026-05-28 (Thu, 28 May 2026)

  Changed paths:
    M flang/lib/Optimizer/Dialect/FIROps.cpp
    M flang/test/Fir/array-coor-canonicalization.fir

  Log Message:
  -----------
  [flang] Canonicalize fir.array_coor for contiguous arrays. (#200106)

This patch adds new canonicalization pattern for `fir.array_coor`
sequences appearing for contiguous array slices like `dx(1:3)`,
`dx(1:3, 2)`, etc. This canonicalization helps exposing the original
rank of the array, which makes MLIR after FIRToMemRef better
optimizable.

Assisted by Cursor



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications