+}
+
+declare <4 x i16> @llvm.bswap.v4i16(<4 x i16>)
+
+define <4 x i16> @test7(<4 x i16> %v) {
+; CHECK-NOSSSE3-LABEL: test7:
+; CHECK-NOSSSE3: # BB#0: # %entry
+; CHECK-NOSSSE3-NEXT: pxor %xmm1, %xmm1
+; CHECK-NOSSSE3-NEXT: movdqa %xmm0, %xmm2
+; CHECK-NOSSSE3-NEXT: punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm1[8],xmm2[9],xmm1[9],xmm2[10],xmm1[10],xmm2[11],xmm1[11],xmm2[12],xmm1[12],xmm2[13],xmm1[13],xmm2[14],xmm1[14],xmm2[15],xmm1[15]
+; CHECK-NOSSSE3-NEXT: pshuflw {{.*#+}} xmm2 = xmm2[3,2,1,0,4,5,6,7]
+; CHECK-NOSSSE3-NEXT: pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,7,6,5,4]
+; CHECK-NOSSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; CHECK-NOSSSE3-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[3,2,1,0,4,5,6,7]
+; CHECK-NOSSSE3-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,6,5,4]
+; CHECK-NOSSSE3-NEXT: packuswb %xmm2, %xmm0
+; CHECK-NOSSSE3-NEXT: psrld $16, %xmm0
+; CHECK-NOSSSE3-NEXT: retq
+;
+; CHECK-SSSE3-LABEL: test7:
+; CHECK-SSSE3: # BB#0: # %entry
+; CHECK-SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[3,2,1,0,7,6,5,4,11,10,9,8,15,14,13,12]
+; CHECK-SSSE3-NEXT: psrld $16, %xmm0
+; CHECK-SSSE3-NEXT: retq
+;
+; CHECK-AVX2-LABEL: test7:
+; CHECK-AVX2: # BB#0: # %entry
+; CHECK-AVX2-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[3,2,1,0,7,6,5,4,11,10,9,8,15,14,13,12]
+; CHECK-AVX2-NEXT: vpsrld $16, %xmm0, %xmm0
+; CHECK-AVX2-NEXT: retq
+;
+; CHECK-WIDE-AVX2-LABEL: test7:
+; CHECK-WIDE-AVX2: # BB#0: # %entry
+; CHECK-WIDE-AVX2-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[1,0,3,2,5,4,7,6,9,8,11,10,13,12,15,14]
+; CHECK-WIDE-AVX2-NEXT: retq
+entry:
+ %r = call <4 x i16> @llvm.bswap.v4i16(<4 x i16> %v)
+ ret <4 x i16> %r
+}
+
+;
+; Double BSWAP -> Identity
+;
+
+define <8 x i16> @identity_v8i16(<8 x i16> %v) {
+; CHECK-ALL-LABEL: identity_v8i16:
+; CHECK-ALL: # BB#0: # %entry
+; CHECK-ALL: retq
+entry:
+ %bs1 = call <8 x i16> @llvm.bswap.v8i16(<8 x i16> %v)
+ %bs2 = call <8 x i16> @llvm.bswap.v8i16(<8 x i16> %bs1)
+ ret <8 x i16> %bs2
+}
+
+define <4 x i32> @identity_v4i32(<4 x i32> %v) {
+; CHECK-ALL-LABEL: identity_v4i32:
+; CHECK-ALL: # BB#0: # %entry
+; CHECK-ALL-NEXT: retq
+entry:
+ %bs1 = call <4 x i32> @llvm.bswap.v4i32(<4 x i32> %v)
+ %bs2 = call <4 x i32> @llvm.bswap.v4i32(<4 x i32> %bs1)
+ ret <4 x i32> %bs2
+}
+
+define <2 x i64> @identity_v2i64(<2 x i64> %v) {
+; CHECK-ALL-LABEL: identity_v2i64:
+; CHECK-ALL: # BB#0: # %entry
+; CHECK-ALL-NEXT: retq
+entry:
+ %bs1 = call <2 x i64> @llvm.bswap.v2i64(<2 x i64> %v)
+ %bs2 = call <2 x i64> @llvm.bswap.v2i64(<2 x i64> %bs1)
+ ret <2 x i64> %bs2
+}
+
+define <16 x i16> @identity_v16i16(<16 x i16> %v) {
+; CHECK-ALL-LABEL: identity_v16i16:
+; CHECK-ALL: # BB#0: # %entry
+; CHECK-ALL-NEXT: retq
+entry:
+ %bs1 = call <16 x i16> @llvm.bswap.v16i16(<16 x i16> %v)
+ %bs2 = call <16 x i16> @llvm.bswap.v16i16(<16 x i16> %bs1)
+ ret <16 x i16> %bs2
+}
+
+define <8 x i32> @identity_v8i32(<8 x i32> %v) {
+; CHECK-ALL-LABEL: identity_v8i32:
+; CHECK-ALL: # BB#0: # %entry
+; CHECK-ALL-NEXT: retq
+entry:
+ %bs1 = call <8 x i32> @llvm.bswap.v8i32(<8 x i32> %v)
+ %bs2 = call <8 x i32> @llvm.bswap.v8i32(<8 x i32> %bs1)
+ ret <8 x i32> %bs2
+}
+
+define <4 x i64> @identity_v4i64(<4 x i64> %v) {
+; CHECK-ALL-LABEL: identity_v4i64:
+; CHECK-ALL: # BB#0: # %entry
+; CHECK-ALL-NEXT: retq
+entry:
+ %bs1 = call <4 x i64> @llvm.bswap.v4i64(<4 x i64> %v)
+ %bs2 = call <4 x i64> @llvm.bswap.v4i64(<4 x i64> %bs1)
+ ret <4 x i64> %bs2
+}
+
+define <4 x i16> @identity_v4i16(<4 x i16> %v) {
+; CHECK-ALL-LABEL: identity_v4i16:
+; CHECK-ALL: # BB#0: # %entry
+; CHECK-ALL-NEXT: retq
+entry:
+ %bs1 = call <4 x i16> @llvm.bswap.v4i16(<4 x i16> %v)
+ %bs2 = call <4 x i16> @llvm.bswap.v4i16(<4 x i16> %bs1)
+ ret <4 x i16> %bs2
+}
+
+;
+; Constant Folding
+;
+
+define <8 x i16> @fold_v8i16() {
+; CHECK-SSE-LABEL: fold_v8i16:
+; CHECK-SSE: # BB#0: # %entry
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm0 = [0,256,65535,512,65023,1024,64511,1536]
+; CHECK-SSE-NEXT: retq
+;
+; CHECK-AVX-LABEL: fold_v8i16:
+; CHECK-AVX: # BB#0: # %entry
+; CHECK-AVX-NEXT: vmovaps {{.*#+}} xmm0 = [0,256,65535,512,65023,1024,64511,1536]
+; CHECK-AVX-NEXT: retq
+entry:
+ %r = call <8 x i16> @llvm.bswap.v8i16(<8 x i16> <i16 0, i16 1, i16 -1, i16 2, i16 -3, i16 4, i16 -5, i16 6>)
+ ret <8 x i16> %r
+}