ret void
}
-; Test the shuffle of a 4xi16 which exposed a problem with the perfect shuffle table
-; entry for vrev.
+; The type <2 x i16> is legalized to <2 x i32> and need to be trunc-stored
+; to <2 x i16> when stored to memory.
define void @test_vrev64(<4 x i16>* nocapture %source, <2 x i16>* nocapture %dst) nounwind ssp {
; CHECK: test_vrev64:
-; CHECK: vrev64.16
-; CHECK: vext.16
+; CHECK: vst1.32
entry:
%0 = bitcast <4 x i16>* %source to <8 x i16>*
%tmp2 = load <8 x i16>* %0, align 4
store <2 x i16> %tmp11, <2 x i16>* %dst, align 4
ret void
}
+
+; Test vrev of float4
+define void @float_vrev64(float* nocapture %source, <4 x float>* nocapture %dest) nounwind noinline ssp {
+; CHECK: float_vrev64
+; CHECK: vext.32
+; CHECK: vrev64.32
+entry:
+ %0 = bitcast float* %source to <4 x float>*
+ %tmp2 = load <4 x float>* %0, align 4
+ %tmp5 = shufflevector <4 x float> <float 0.000000e+00, float undef, float undef, float undef>, <4 x float> %tmp2, <4 x i32> <i32 0, i32 7, i32 0, i32 0>
+ %arrayidx8 = getelementptr inbounds <4 x float>* %dest, i32 11
+ store <4 x float> %tmp5, <4 x float>* %arrayidx8, align 4
+ ret void
+}
+