[llvm] [AMDGPU] Use reverse iteration in CodeGenPrepare (PR #145484)

[Jay Foad via llvm-commits]

================
@@ -1305,7 +1298,8 @@ within the byte are all 0.
 static bool tryNarrowMathIfNoOverflow(Instruction *I,
                                       const SITargetLowering *TLI,
                                       const TargetTransformInfo &TTI,
-                                      const DataLayout &DL) {
+                                      const DataLayout &DL,
+                                      SmallVector<WeakVH> &DeadVals) {
----------------
jayfoad wrote:

As a general rule I would prefer making this a member function, but I don't know if it's worth it in this case.

https://github.com/llvm/llvm-project/pull/145484