[Mlir-commits] [clang] [llvm] [mlir] [IR][CodeGen] Replace constrained FP intrinsics with fp.control/fp.except operand bundles (PR #191613)

[Jason Eckhardt]

================
@@ -0,0 +1,121 @@
+//===- llvm/IR/FloatingPointOps.def - FP intrinsics -------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+// Defines the set of intrinsics classified as floating-point operations.
+//
+// Two macro forms:
+//
+//   FP_OP(intrinsic_name, isd_opcode_name)
+//     - An FP intrinsic that lowers to a single ISD opcode.
+//     - isFloatingPointOperation() uses column 1 (Intrinsic::NAME).
+//     - ISD::isFPOpcode() uses column 2 (ISD::ISD_NAME).
+//
+//   FP_INTRINSIC(intrinsic_name)
+//     - An FP intrinsic whose ISD opcode is already covered by another FP_OP
+//       entry, or maps to a shared opcode (e.g. SETCC) not exclusively FP.
+//     - isFloatingPointOperation() uses it; ISD::isFPOpcode() does not.
+//
+//===----------------------------------------------------------------------===//
----------------
nvjle wrote:

CS nit: Can we use the updated header style? https://llvm.org/docs/CodingStandards.html#file-headers

We don't need the file information in the header line anymore.

https://github.com/llvm/llvm-project/pull/191613