[llvm] [AMDGPU] Disallow null for more resource operands (PR #121941)

[Jay Foad via llvm-commits]

================
@@ -680,7 +680,7 @@ multiclass MUBUF_Pseudo_Stores<string opName, ValueType store_vt = i32> {
 class MUBUF_Pseudo_Store_Lds<string opName>
   : MUBUF_Pseudo<opName,
                  (outs),
-                 (ins SReg_128:$srsrc, SCSrc_b32:$soffset, Offset:$offset, CPol:$cpol, i1imm:$swz),
+                 (ins SReg_128_XNULL:$srsrc, SCSrc_b32:$soffset, Offset:$offset, CPol:$cpol, i1imm:$swz),
----------------
jayfoad wrote:

This change is just for consistency. It doesn't make a practical difference because this instruction only exists pre-GFX10, but "null" only exists in GFX10+.

https://github.com/llvm/llvm-project/pull/121941