[llvm] MCExpr-ify AMDGPU PALMetadata (PR #93236)

[Matt Arsenault via llvm-commits]

================
@@ -0,0 +1,79 @@
+//===-- SIDefines.h - SI Helper Functions -----------------------*- 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
+//
+/// \file - utility functions for the SIDefines and its common uses.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIB_TARGET_AMDGPU_UTILS_SIDEFINESUTILS_H
+#define LLVM_LIB_TARGET_AMDGPU_UTILS_SIDEFINESUTILS_H
+
+#include "llvm/MC/MCExpr.h"
+#include <utility>
+
+namespace llvm {
+class MCContext;
+namespace AMDGPU {
+
+/// Deduce the least significant bit aligned shift and mask values for a binary
+/// Complement \p Value (as they're defined in SIDefines.h as C_*) as a returned
+/// pair<shift, mask>. That is to say \p Value == ~(mask << shift)
+///
+/// For example, given C_00B848_FWD_PROGRESS (i.e., 0x7FFFFFFF) from
+/// SIDefines.h, this will return the pair as (31,1).
+constexpr inline std::pair<unsigned, unsigned> getShiftMask(unsigned Value) {
----------------
arsenm wrote:

inline is redundant with constexpr 

https://github.com/llvm/llvm-project/pull/93236