Add impl for llvm.roundeven in cg_clif

- remove unused `llvm.aarch64.neon.frintn` from cg_clif

compiler/rustc_codegen_cranelift/example/neon.rs

@@ -233,7 +233,7 @@ unsafe fn test_vaddvq_f32() {
 
 #[cfg(target_arch = "aarch64")]
 unsafe fn test_vrndnq_f32() {
-    // AArch64 llvm intrinsic: llvm.aarch64.neon.frintn.v4f32
+    // llvm intrinsic: llvm.roundeven.v4f32
     let a = f32x4::from([0.1, -1.9, 4.5, 5.5]);
     let e = f32x4::from([0., -2., 4., 6.]);
     let r: f32x4 = transmute(vrndnq_f32(transmute(a)));

compiler/rustc_codegen_cranelift/src/intrinsics/llvm.rs

@@ -62,6 +62,14 @@ pub(crate) fn codegen_llvm_intrinsic_call<'tcx>(
             });
         }
 
+        _ if intrinsic.starts_with("llvm.roundeven.v") => {
+            intrinsic_args!(fx, args => (v); intrinsic);
+
+            simd_for_each_lane(fx, v, ret, &|fx, _lane_ty, _res_lane_ty, lane| {
+                fx.bcx.ins().nearest(lane)
+            });
+        }
+
         _ => {
             fx.tcx
                 .dcx()

compiler/rustc_codegen_cranelift/src/intrinsics/llvm_aarch64.rs

@@ -264,14 +264,6 @@ pub(super) fn codegen_aarch64_llvm_intrinsic_call<'tcx>(
             simd_reduce(fx, v, None, ret, &|fx, _ty, a, b| fx.bcx.ins().fadd(a, b));
         }
 
-        _ if intrinsic.starts_with("llvm.aarch64.neon.frintn.v") => {
-            intrinsic_args!(fx, args => (v); intrinsic);
-
-            simd_for_each_lane(fx, v, ret, &|fx, _lane_ty, _res_lane_ty, lane| {
-                fx.bcx.ins().nearest(lane)
-            });
-        }
-
         _ if intrinsic.starts_with("llvm.aarch64.neon.smaxv.i") => {
             intrinsic_args!(fx, args => (v); intrinsic);