[RISCV] Implement Builtins for XAndesBFHCvt extension. (#148804)

XAndesBFHCvt provides two builtins functions for converting between
float and bf16. Users can use them to convert bf16 values loaded from
memory to float, perform arithmetic operations, then convert them back
to bf16 and store them to memory.

The load/store and move operations for bf16 will be handled in a later
patch.

1 files changed, 6 insertions, 0 deletions

diff --git a/clang/lib/CodeGen/TargetBuiltins/RISCV.cpp b/clang/lib/CodeGen/TargetBuiltins/RISCV.cpp
index 89e3f6f..b08a058 100644
--- a/clang/lib/CodeGen/TargetBuiltins/RISCV.cpp
+++ b/clang/lib/CodeGen/TargetBuiltins/RISCV.cpp
@@ -413,6 +413,12 @@ Value *CodeGenFunction::EmitRISCVBuiltinExpr(unsigned BuiltinID,
     ID = Intrinsic::riscv_cv_alu_subuRN;
     break;
 
+  // XAndesBFHCvt
+  case RISCV::BI__builtin_riscv_nds_fcvt_s_bf16:
+    return Builder.CreateFPExt(Ops[0], FloatTy);
+  case RISCV::BI__builtin_riscv_nds_fcvt_bf16_s:
+    return Builder.CreateFPTrunc(Ops[0], BFloatTy);
+
     // Vector builtins are handled from here.
 #include "clang/Basic/riscv_vector_builtin_cg.inc"