[llvm] [AMDGPU][True16][MC][CodeGen] true16 for v_cndmask_b16 (PR #119736)

[Joe Nash via llvm-commits]

================
@@ -765,8 +785,8 @@ def VOP_WRITELANE : VOPProfile<[i32, i32, i32, i32]> {
 // VOP2 Instructions
 //===----------------------------------------------------------------------===//
 
-let SubtargetPredicate = isGFX11Plus in
-defm V_CNDMASK_B16 : VOP2eInst <"v_cndmask_b16", VOP2e_I16_I16_I16_I1_fake16>;
+defm V_CNDMASK_B16_t16 : VOP2eInst <"v_cndmask_b16_t16", VOP2e_I16_I16_I16_I1_true16>;
----------------
Sisyph wrote:

We should have SubtargetPredicate and True16Predicate on these instructions. I guess the codegen works since the selection patterns have the predicates, but MC probably needs them.

https://github.com/llvm/llvm-project/pull/119736