[llvm-branch-commits] [llvm] [SelectionDAG][X86] Split via Concat <n x T> vector types for atomic load (PR #120640)

[via llvm-branch-commits]

================
@@ -194,8 +194,8 @@ bool BaseIndexOffset::contains(const SelectionDAG &DAG, int64_t BitSize,
   return false;
 }
 
-/// Parses tree in Ptr for base, index, offset addresses.
-static BaseIndexOffset matchLSNode(const LSBaseSDNode *N,
+template <typename T>
+static BaseIndexOffset matchSDNode(const T *N,
----------------
jofrn wrote:

Without this, the split elements are not identified, and we end up having codegen emit extra moves for each element even though they are not used.

https://github.com/llvm/llvm-project/pull/120640