[all-commits] [llvm/llvm-project] 0b3943: [MLIR][SCF] fix scf.index_switch fold convergence ...

[Keyi Zhang via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 0b3943f3bba71e0cf9ea9984dae472cf503bca21
      https://github.com/llvm/llvm-project/commit/0b3943f3bba71e0cf9ea9984dae472cf503bca21
  Author: Keyi Zhang <Kuree at users.noreply.github.com>
  Date:   2024-07-16 (Tue, 16 Jul 2024)

  Changed paths:
    M mlir/include/mlir/Dialect/SCF/IR/SCFOps.td
    M mlir/lib/Dialect/SCF/IR/SCF.cpp
    M mlir/test/Dialect/SCF/canonicalize.mlir

  Log Message:
  -----------
  [MLIR][SCF] fix scf.index_switch fold convergence (#98535) (#98680)

If the `scf.index_switch` op has no result, the current fold logic
results in an infinite loop (see #98535). The is because `fold`
mechanism does not support *erasing* zero-result ops. This PR moves the
fold logic to a canonicalizer and fix the issue.



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