[llvm] [AMDGPU][MC] Allow null where 128b or larger dst reg is expected (PR #115200)

[Matt Arsenault via llvm-commits]

================
@@ -239,6 +239,22 @@ s_load_b512 s[20:35], s[2:3], s0 glc dlc
 s_load_b512 s[20:35], s[2:3], 0x1234 glc dlc
 // GFX11: encoding: [0x01,0x65,0x10,0xf4,0x34,0x12,0x00,0xf8]
 
+// null as dst
+s_load_b32 null, s[2:3], s0
+// GFX11: encoding: [0x01,0x1f,0x00,0xf4,0x00,0x00,0x00,0x00]
+
+s_load_b64 null, s[2:3], s0
+// GFX11: encoding: [0x01,0x1f,0x04,0xf4,0x00,0x00,0x00,0x00]
+
----------------
arsenm wrote:

Test the b96 case 

https://github.com/llvm/llvm-project/pull/115200