[llvm] [AMDGPU] Fix opsel for scaled MFMA operations (PR #140183)

[Matt Arsenault via llvm-commits]

================
@@ -526,14 +526,14 @@ class VOP3PXe <bits<7> op, VOPProfile MFMAPfl, bit acc_cd = 0> : Enc128, VOP3Pe_
   bits<9> scale_src0;
   bits<9> scale_src1;
 
-  bits<2> scale_src0_opsel;
-  bits<2> scale_src1_opsel;
+  bits<4> src0_modifiers;
+  bits<4> src1_modifiers;
----------------
arsenm wrote:

Needs comment explaining these. If you're writing custom operand handling maybe we shouldn't rename these? It is a weird case 

https://github.com/llvm/llvm-project/pull/140183