[all-commits] [llvm/llvm-project] fcd06d: [mlir][flang] add fast math attribute to fcmp (#74...

[Tom Eccles via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: fcd06d774dad1b5bcdc22f5f10a9d5c62c585814
      https://github.com/llvm/llvm-project/commit/fcd06d774dad1b5bcdc22f5f10a9d5c62c585814
  Author: Tom Eccles <tom.eccles at arm.com>
  Date:   2023-12-06 (Wed, 06 Dec 2023)

  Changed paths:
    M flang/include/flang/Optimizer/Dialect/FIROps.td
    M flang/lib/Optimizer/CodeGen/CodeGen.cpp
    M flang/test/Fir/convert-to-llvm.fir
    M flang/test/Fir/fir-ops.fir
    M flang/test/HLFIR/call_with_poly_dummy.f90
    M flang/test/Lower/HLFIR/binary-ops.f90
    M flang/test/Lower/HLFIR/user-defined-assignment.f90
    M flang/test/Lower/Intrinsics/bessel_jn.f90
    M flang/test/Lower/Intrinsics/bessel_yn.f90
    M flang/test/Lower/Intrinsics/dim.f90
    M flang/test/Lower/Intrinsics/min.f90
    M flang/test/Lower/Intrinsics/modulo.f90
    M flang/test/Lower/Intrinsics/nearest.f90
    M flang/test/Lower/OpenACC/acc-reduction.f90
    M flang/test/Lower/arithmetic-goto.f90
    M flang/test/Lower/array-elemental-calls-2.f90
    M flang/test/Lower/array-expression-slice-1.f90
    M flang/test/Lower/array-user-def-assignments.f90
    M flang/test/Lower/forall/forall-construct.f90
    M flang/test/Lower/forall/forall-where.f90
    M flang/test/Lower/host-associated.f90
    M flang/test/Lower/real-operations-2.f90
    M flang/test/Lower/where.f90
    M flang/test/Transforms/simplifyintrinsics.fir
    M mlir/include/mlir/Dialect/Arith/IR/ArithOps.td
    M mlir/lib/Conversion/ArithToLLVM/ArithToLLVM.cpp
    M mlir/test/Conversion/ArithToLLVM/arith-to-llvm.mlir
    M mlir/test/Dialect/Arith/ops.mlir

  Log Message:
  -----------
  [mlir][flang] add fast math attribute to fcmp (#74315)

`llvm.fcmp` does support fast math attributes therefore so should
`arith.cmpf`.

The heavy churn in flang tests are because flang sets
`fastmath<contract>` by default on all operations that support the fast
math interface. Downstream users of MLIR should not be so effected.

This was requested in https://github.com/llvm/llvm-project/issues/74263