[x86] Teach the unpack lowering to try wider element unpacks.

[oota-llvm.git] / test / CodeGen / X86 / vector-shuffle-combining.ll

diff --git a/test/CodeGen/X86/vector-shuffle-combining.ll b/test/CodeGen/X86/vector-shuffle-combining.ll
index b870c366181c09cd0ba6f39ba160b548d8e13672..a1051c6610d0825c648903d6c7417ce78743cb24 100644 (file)
--- a/test/CodeGen/X86/vector-shuffle-combining.ll
+++ b/test/CodeGen/X86/vector-shuffle-combining.ll
@@ -275,16 +275,18 @@ define <4 x i32> @combine_bitwise_ops_test6(<4 x i32> %a, <4 x i32> %b, <4 x i32
 define <4 x i32> @combine_bitwise_ops_test1b(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c) {
 ; SSE2-LABEL: combine_bitwise_ops_test1b:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    andps %xmm1, %xmm0
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,3,2,3]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_bitwise_ops_test1b:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    andps %xmm1, %xmm0
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; SSSE3-NEXT:    pand %xmm1, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,3,2,3]
+; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_bitwise_ops_test1b:
@@ -313,16 +315,18 @@ define <4 x i32> @combine_bitwise_ops_test1b(<4 x i32> %a, <4 x i32> %b, <4 x i3
 define <4 x i32> @combine_bitwise_ops_test2b(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c) {
 ; SSE2-LABEL: combine_bitwise_ops_test2b:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    orps %xmm1, %xmm0
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,3,2,3]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_bitwise_ops_test2b:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    orps %xmm1, %xmm0
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; SSSE3-NEXT:    por %xmm1, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,3,2,3]
+; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_bitwise_ops_test2b:
@@ -390,18 +394,18 @@ define <4 x i32> @combine_bitwise_ops_test3b(<4 x i32> %a, <4 x i32> %b, <4 x i3
 define <4 x i32> @combine_bitwise_ops_test4b(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c) {
 ; SSE2-LABEL: combine_bitwise_ops_test4b:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    andps %xmm1, %xmm0
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; SSE2-NEXT:    movaps %xmm2, %xmm0
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,3,2,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_bitwise_ops_test4b:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    andps %xmm1, %xmm0
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; SSSE3-NEXT:    movaps %xmm2, %xmm0
+; SSSE3-NEXT:    pand %xmm1, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,3,2,3]
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_bitwise_ops_test4b:
@@ -430,18 +434,18 @@ define <4 x i32> @combine_bitwise_ops_test4b(<4 x i32> %a, <4 x i32> %b, <4 x i3
 define <4 x i32> @combine_bitwise_ops_test5b(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c) {
 ; SSE2-LABEL: combine_bitwise_ops_test5b:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    orps %xmm1, %xmm0
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; SSE2-NEXT:    movaps %xmm2, %xmm0
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,3,2,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_bitwise_ops_test5b:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    orps %xmm1, %xmm0
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; SSSE3-NEXT:    movaps %xmm2, %xmm0
+; SSSE3-NEXT:    por %xmm1, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,3,2,3]
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_bitwise_ops_test5b:
@@ -507,17 +511,42 @@ define <4 x i32> @combine_bitwise_ops_test6b(<4 x i32> %a, <4 x i32> %b, <4 x i3
 }
 
 define <4 x i32> @combine_bitwise_ops_test1c(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c) {
-; SSE-LABEL: combine_bitwise_ops_test1c:
-; SSE:       # BB#0:
-; SSE-NEXT:    andps %xmm1, %xmm0
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[1,3]
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_bitwise_ops_test1c:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,3,2,3]
+; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: combine_bitwise_ops_test1c:
-; AVX:       # BB#0:
-; AVX-NEXT:    vandps %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[1,3]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: combine_bitwise_ops_test1c:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    pand %xmm1, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,3,2,3]
+; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_bitwise_ops_test1c:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pand %xmm1, %xmm0
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,3],xmm0[4,5],xmm2[6,7]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: combine_bitwise_ops_test1c:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,3],xmm0[4,5],xmm2[6,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_bitwise_ops_test1c:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm2[1],xmm0[2],xmm2[3]
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX2-NEXT:    retq
   %shuf1 = shufflevector <4 x i32> %a, <4 x i32> %c, <4 x i32><i32 0, i32 2, i32 5, i32 7>
   %shuf2 = shufflevector <4 x i32> %b, <4 x i32> %c, <4 x i32><i32 0, i32 2, i32 5, i32 7>
   %and = and <4 x i32> %shuf1, %shuf2
@@ -525,17 +554,42 @@ define <4 x i32> @combine_bitwise_ops_test1c(<4 x i32> %a, <4 x i32> %b, <4 x i3
 }
 
 define <4 x i32> @combine_bitwise_ops_test2c(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c) {
-; SSE-LABEL: combine_bitwise_ops_test2c:
-; SSE:       # BB#0:
-; SSE-NEXT:    orps %xmm1, %xmm0
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[1,3]
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_bitwise_ops_test2c:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,3,2,3]
+; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: combine_bitwise_ops_test2c:
-; AVX:       # BB#0:
-; AVX-NEXT:    vorps %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[1,3]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: combine_bitwise_ops_test2c:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    por %xmm1, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,3,2,3]
+; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_bitwise_ops_test2c:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    por %xmm1, %xmm0
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,3],xmm0[4,5],xmm2[6,7]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: combine_bitwise_ops_test2c:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,3],xmm0[4,5],xmm2[6,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_bitwise_ops_test2c:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpor %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm2[1],xmm0[2],xmm2[3]
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX2-NEXT:    retq
   %shuf1 = shufflevector <4 x i32> %a, <4 x i32> %c, <4 x i32><i32 0, i32 2, i32 5, i32 7>
   %shuf2 = shufflevector <4 x i32> %b, <4 x i32> %c, <4 x i32><i32 0, i32 2, i32 5, i32 7>
   %or = or <4 x i32> %shuf1, %shuf2
@@ -545,28 +599,32 @@ define <4 x i32> @combine_bitwise_ops_test2c(<4 x i32> %a, <4 x i32> %b, <4 x i3
 define <4 x i32> @combine_bitwise_ops_test3c(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c) {
 ; SSE2-LABEL: combine_bitwise_ops_test3c:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    xorps %xmm1, %xmm0
-; SSE2-NEXT:    xorps %xmm1, %xmm1
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,3]
+; SSE2-NEXT:    pxor %xmm1, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SSE2-NEXT:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_bitwise_ops_test3c:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    xorps %xmm1, %xmm0
-; SSSE3-NEXT:    xorps %xmm1, %xmm1
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,3]
+; SSSE3-NEXT:    pxor %xmm1, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SSSE3-NEXT:    pxor %xmm1, %xmm1
+; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_bitwise_ops_test3c:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    xorps %xmm1, %xmm0
-; SSE41-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0,2],zero,zero
+; SSE41-NEXT:    pxor %xmm1, %xmm0
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SSE41-NEXT:    movq {{.*#+}} xmm0 = xmm0[0],zero
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_bitwise_ops_test3c:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vxorps %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0,2],zero,zero
+; AVX-NEXT:    vpxor %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX-NEXT:    vmovq {{.*#+}} xmm0 = xmm0[0],zero
 ; AVX-NEXT:    retq
   %shuf1 = shufflevector <4 x i32> %a, <4 x i32> %c, <4 x i32><i32 0, i32 2, i32 5, i32 7>
   %shuf2 = shufflevector <4 x i32> %b, <4 x i32> %c, <4 x i32><i32 0, i32 2, i32 5, i32 7>
@@ -575,18 +633,42 @@ define <4 x i32> @combine_bitwise_ops_test3c(<4 x i32> %a, <4 x i32> %b, <4 x i3
 }
 
 define <4 x i32> @combine_bitwise_ops_test4c(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c) {
-; SSE-LABEL: combine_bitwise_ops_test4c:
-; SSE:       # BB#0:
-; SSE-NEXT:    andps %xmm1, %xmm0
-; SSE-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSE-NEXT:    movaps %xmm2, %xmm0
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_bitwise_ops_test4c:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,3,2,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: combine_bitwise_ops_test4c:
-; AVX:       # BB#0:
-; AVX-NEXT:    vandps %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm2[0,2],xmm0[1,3]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: combine_bitwise_ops_test4c:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    pand %xmm1, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,3,2,3]
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_bitwise_ops_test4c:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pand %xmm1, %xmm0
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm2[0,1],xmm0[2,3],xmm2[4,5],xmm0[6,7]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: combine_bitwise_ops_test4c:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm2[0,1],xmm0[2,3],xmm2[4,5],xmm0[6,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_bitwise_ops_test4c:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm2[0],xmm0[1],xmm2[2],xmm0[3]
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX2-NEXT:    retq
   %shuf1 = shufflevector <4 x i32> %c, <4 x i32> %a, <4 x i32><i32 0, i32 2, i32 5, i32 7>
   %shuf2 = shufflevector <4 x i32> %c, <4 x i32> %b, <4 x i32><i32 0, i32 2, i32 5, i32 7>
   %and = and <4 x i32> %shuf1, %shuf2
@@ -594,18 +676,42 @@ define <4 x i32> @combine_bitwise_ops_test4c(<4 x i32> %a, <4 x i32> %b, <4 x i3
 }
 
 define <4 x i32> @combine_bitwise_ops_test5c(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c) {
-; SSE-LABEL: combine_bitwise_ops_test5c:
-; SSE:       # BB#0:
-; SSE-NEXT:    orps %xmm1, %xmm0
-; SSE-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSE-NEXT:    movaps %xmm2, %xmm0
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_bitwise_ops_test5c:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,3,2,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: combine_bitwise_ops_test5c:
-; AVX:       # BB#0:
-; AVX-NEXT:    vorps %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm2[0,2],xmm0[1,3]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: combine_bitwise_ops_test5c:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    por %xmm1, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,3,2,3]
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_bitwise_ops_test5c:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    por %xmm1, %xmm0
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm2[0,1],xmm0[2,3],xmm2[4,5],xmm0[6,7]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: combine_bitwise_ops_test5c:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm2[0,1],xmm0[2,3],xmm2[4,5],xmm0[6,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_bitwise_ops_test5c:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpor %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm2[0],xmm0[1],xmm2[2],xmm0[3]
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX2-NEXT:    retq
   %shuf1 = shufflevector <4 x i32> %c, <4 x i32> %a, <4 x i32><i32 0, i32 2, i32 5, i32 7>
   %shuf2 = shufflevector <4 x i32> %c, <4 x i32> %b, <4 x i32><i32 0, i32 2, i32 5, i32 7>
   %or = or <4 x i32> %shuf1, %shuf2
@@ -613,20 +719,45 @@ define <4 x i32> @combine_bitwise_ops_test5c(<4 x i32> %a, <4 x i32> %b, <4 x i3
 }
 
 define <4 x i32> @combine_bitwise_ops_test6c(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c) {
-; SSE-LABEL: combine_bitwise_ops_test6c:
-; SSE:       # BB#0:
-; SSE-NEXT:    xorps %xmm1, %xmm0
-; SSE-NEXT:    xorps %xmm1, %xmm1
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[1,3]
-; SSE-NEXT:    movaps %xmm1, %xmm0
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_bitwise_ops_test6c:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    pxor %xmm1, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,3,2,3]
+; SSE2-NEXT:    pxor %xmm0, %xmm0
+; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: combine_bitwise_ops_test6c:
-; AVX:       # BB#0:
-; AVX-NEXT:    vxorps %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    vxorps %xmm1, %xmm1, %xmm1
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[0,2],xmm0[1,3]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: combine_bitwise_ops_test6c:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    pxor %xmm1, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,3,2,3]
+; SSSE3-NEXT:    pxor %xmm0, %xmm0
+; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_bitwise_ops_test6c:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pxor %xmm1, %xmm0
+; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[0,1,1,3]
+; SSE41-NEXT:    pxor %xmm0, %xmm0
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: combine_bitwise_ops_test6c:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpxor %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,1,3]
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1,2,3],xmm0[4,5,6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_bitwise_ops_test6c:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpxor %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,1,3]
+; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3]
+; AVX2-NEXT:    retq
   %shuf1 = shufflevector <4 x i32> %c, <4 x i32> %a, <4 x i32><i32 0, i32 2, i32 5, i32 7>
   %shuf2 = shufflevector <4 x i32> %c, <4 x i32> %b, <4 x i32><i32 0, i32 2, i32 5, i32 7>
   %xor = xor <4 x i32> %shuf1, %shuf2
@@ -857,52 +988,112 @@ define <4 x i32> @combine_nested_undef_test14(<4 x i32> %A, <4 x i32> %B) {
 ; it.
 
 define <4 x i32> @combine_nested_undef_test15(<4 x i32> %A, <4 x i32> %B) {
-; SSE-LABEL: combine_nested_undef_test15:
-; SSE:       # BB#0:
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[3,0]
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[0,1]
-; SSE-NEXT:    movaps %xmm1, %xmm0
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_nested_undef_test15:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[3,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[0,1]
+; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: combine_nested_undef_test15:
-; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[3,0]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[2,0],xmm0[0,1]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: combine_nested_undef_test15:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[3,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[0,1]
+; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_nested_undef_test15:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,0,1]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: combine_nested_undef_test15:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,1,0,1]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_nested_undef_test15:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpbroadcastd %xmm1, %xmm1
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,1,0,1]
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3]
+; AVX2-NEXT:    retq
   %1 = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 0, i32 4, i32 3, i32 1>
   %2 = shufflevector <4 x i32> %1, <4 x i32> undef, <4 x i32> <i32 2, i32 1, i32 0, i32 3>
   ret <4 x i32> %2
 }
 
 define <4 x i32> @combine_nested_undef_test16(<4 x i32> %A, <4 x i32> %B) {
-; SSE-LABEL: combine_nested_undef_test16:
-; SSE:       # BB#0:
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[1,3]
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_nested_undef_test16:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,0,2,3]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: combine_nested_undef_test16:
-; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[1,3]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: combine_nested_undef_test16:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3]
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,0,2,3]
+; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_nested_undef_test16:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5],xmm1[6,7]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: combine_nested_undef_test16:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5],xmm1[6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_nested_undef_test16:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
+; AVX2-NEXT:    retq
   %1 = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
   %2 = shufflevector <4 x i32> %1, <4 x i32> undef, <4 x i32> <i32 2, i32 1, i32 0, i32 3>
   ret <4 x i32> %2
 }
 
 define <4 x i32> @combine_nested_undef_test17(<4 x i32> %A, <4 x i32> %B) {
-; SSE-LABEL: combine_nested_undef_test17:
-; SSE:       # BB#0:
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,0]
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,1],xmm1[0,2]
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_nested_undef_test17:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,1],xmm1[0,2]
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: combine_nested_undef_test17:
-; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,0]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[3,1],xmm1[0,2]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: combine_nested_undef_test17:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,1],xmm1[0,2]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_nested_undef_test17:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3,4,5,6,7]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,0,1]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: combine_nested_undef_test17:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3,4,5,6,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,1,0,1]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_nested_undef_test17:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm1[0],xmm0[1,2,3]
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,1,0,1]
+; AVX2-NEXT:    retq
   %1 = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 4, i32 1, i32 3, i32 1>
   %2 = shufflevector <4 x i32> %1, <4 x i32> undef, <4 x i32> <i32 2, i32 1, i32 0, i32 3>
   ret <4 x i32> %2
@@ -924,54 +1115,111 @@ define <4 x i32> @combine_nested_undef_test18(<4 x i32> %A, <4 x i32> %B) {
 }
 
 define <4 x i32> @combine_nested_undef_test19(<4 x i32> %A, <4 x i32> %B) {
-; SSE-LABEL: combine_nested_undef_test19:
-; SSE:       # BB#0:
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,0],xmm0[0,0]
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[0,0]
-; SSE-NEXT:    movaps %xmm1, %xmm0
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_nested_undef_test19:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,0],xmm0[0,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[0,0]
+; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: combine_nested_undef_test19:
-; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[1,0],xmm0[0,0]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[0,2],xmm0[0,0]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: combine_nested_undef_test19:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,0],xmm0[0,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[0,0]
+; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_nested_undef_test19:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,0,0,0]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: combine_nested_undef_test19:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,0,0,0]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_nested_undef_test19:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3]
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,0,0,0]
+; AVX2-NEXT:    retq
   %1 = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 0, i32 4, i32 5, i32 6>
   %2 = shufflevector <4 x i32> %1, <4 x i32> undef, <4 x i32> <i32 2, i32 0, i32 0, i32 0>
   ret <4 x i32> %2
 }
 
 define <4 x i32> @combine_nested_undef_test20(<4 x i32> %A, <4 x i32> %B) {
-; SSE-LABEL: combine_nested_undef_test20:
-; SSE:       # BB#0:
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[2,3]
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2,3,1]
-; SSE-NEXT:    movaps %xmm1, %xmm0
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_nested_undef_test20:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[2,3]
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2,3,1]
+; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: combine_nested_undef_test20:
-; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[0,0],xmm0[2,3]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2,3,1]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: combine_nested_undef_test20:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[2,3]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2,3,1]
+; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_nested_undef_test20:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm1[0,1,2,3],xmm0[4,5,6,7]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,3,0]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: combine_nested_undef_test20:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1,2,3],xmm0[4,5,6,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,3,0]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_nested_undef_test20:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3]
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,3,0]
+; AVX2-NEXT:    retq
   %1 = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 3, i32 2, i32 4, i32 4>
   %2 = shufflevector <4 x i32> %1, <4 x i32> undef, <4 x i32> <i32 2, i32 1, i32 0, i32 3>
   ret <4 x i32> %2
 }
 
 define <4 x i32> @combine_nested_undef_test21(<4 x i32> %A, <4 x i32> %B) {
-; SSE-LABEL: combine_nested_undef_test21:
-; SSE:       # BB#0:
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,1]
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2,1,3]
-; SSE-NEXT:    movaps %xmm1, %xmm0
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_nested_undef_test21:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,0,1,1]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: combine_nested_undef_test21:
-; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[0,0],xmm0[1,1]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: combine_nested_undef_test21:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,0,1,1]
+; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_nested_undef_test21:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3],xmm1[4,5,6,7]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: combine_nested_undef_test21:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3],xmm1[4,5,6,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_nested_undef_test21:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2,3]
+; AVX2-NEXT:    vpbroadcastq %xmm0, %xmm0
+; AVX2-NEXT:    retq
   %1 = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 4, i32 1, i32 3, i32 1>
   %2 = shufflevector <4 x i32> %1, <4 x i32> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 3>
   ret <4 x i32> %2
@@ -1599,16 +1847,28 @@ define <4 x float> @combine_test2b(<4 x float> %a, <4 x float> %b) {
 }
 
 define <4 x float> @combine_test3b(<4 x float> %a, <4 x float> %b) {
-; SSE-LABEL: combine_test3b:
-; SSE:       # BB#0:
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[3,0]
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[2,3]
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_test3b:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[3,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[2,3]
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: combine_test3b:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[3,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[2,3]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_test3b:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    blendpd {{.*#+}} xmm0 = xmm0[0],xmm1[1]
+; SSE41-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,3,2,3]
+; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_test3b:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[3,0]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[2,3]
+; AVX-NEXT:    vblendpd {{.*#+}} xmm0 = xmm0[0],xmm1[1]
+; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,3,2,3]
 ; AVX-NEXT:    retq
   %1 = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 0, i32 0, i32 6, i32 3>
   %2 = shufflevector <4 x float> %1, <4 x float> %b, <4 x i32> <i32 0, i32 7, i32 2, i32 7>
@@ -2537,3 +2797,115 @@ define <4 x float> @combine_insertps4(<4 x float> %a, <4 x float> %b) {
   %d = shufflevector <4 x float> %a, <4 x float> %c, <4 x i32><i32 4, i32 1, i32 6, i32 5>
   ret <4 x float> %d
 }
+
+define <4 x float> @PR22377(<4 x float> %a, <4 x float> %b) {
+; SSE-LABEL: PR22377:
+; SSE:       # BB#0: # %entry
+; SSE-NEXT:    movaps %xmm0, %xmm1
+; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,3,1,3]
+; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,0,2]
+; SSE-NEXT:    addps %xmm0, %xmm1
+; SSE-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE-NEXT:    retq
+;
+; AVX-LABEL: PR22377:
+; AVX:       # BB#0: # %entry
+; AVX-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[1,3,1,3]
+; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,0,2]
+; AVX-NEXT:    vaddps %xmm0, %xmm1, %xmm1
+; AVX-NEXT:    vunpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; AVX-NEXT:    retq
+entry:
+  %s1 = shufflevector <4 x float> %a, <4 x float> undef, <4 x i32> <i32 1, i32 3, i32 1, i32 3>
+  %s2 = shufflevector <4 x float> %a, <4 x float> undef, <4 x i32> <i32 0, i32 2, i32 0, i32 2>
+  %r2 = fadd <4 x float> %s1, %s2
+  %s3 = shufflevector <4 x float> %s2, <4 x float> %r2, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
+  ret <4 x float> %s3
+}
+
+define <4 x float> @PR22390(<4 x float> %a, <4 x float> %b) {
+; SSE2-LABEL: PR22390:
+; SSE2:       # BB#0: # %entry
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,0,1,2]
+; SSE2-NEXT:    movaps %xmm0, %xmm2
+; SSE2-NEXT:    movss {{.*#+}} xmm2 = xmm1[0],xmm2[1,2,3]
+; SSE2-NEXT:    addps %xmm0, %xmm2
+; SSE2-NEXT:    movaps %xmm2, %xmm0
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: PR22390:
+; SSSE3:       # BB#0: # %entry
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,0,1,2]
+; SSSE3-NEXT:    movaps %xmm0, %xmm2
+; SSSE3-NEXT:    movss {{.*#+}} xmm2 = xmm1[0],xmm2[1,2,3]
+; SSSE3-NEXT:    addps %xmm0, %xmm2
+; SSSE3-NEXT:    movaps %xmm2, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: PR22390:
+; SSE41:       # BB#0: # %entry
+; SSE41-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,0,1,2]
+; SSE41-NEXT:    blendps {{.*#+}} xmm1 = xmm1[0],xmm0[1,2,3]
+; SSE41-NEXT:    addps %xmm1, %xmm0
+; SSE41-NEXT:    retq
+;
+; AVX-LABEL: PR22390:
+; AVX:       # BB#0: # %entry
+; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[3,0,1,2]
+; AVX-NEXT:    vblendps {{.*#+}} xmm1 = xmm1[0],xmm0[1,2,3]
+; AVX-NEXT:    vaddps %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    retq
+entry:
+  %s1 = shufflevector <4 x float> %a, <4 x float> undef, <4 x i32> <i32 3, i32 0, i32 1, i32 2>
+  %s2 = shufflevector <4 x float> %s1, <4 x float> %b, <4 x i32> <i32 4, i32 1, i32 2, i32 3>
+  %r2 = fadd <4 x float> %s1, %s2
+  ret <4 x float> %r2
+}
+
+define <8 x float> @PR22412(<8 x float> %a, <8 x float> %b) {
+; SSE2-LABEL: PR22412:
+; SSE2:       # BB#0: # %entry
+; SSE2-NEXT:    movsd {{.*#+}} xmm2 = xmm0[0],xmm2[1]
+; SSE2-NEXT:    movapd %xmm2, %xmm0
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm3[3,2]
+; SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,0],xmm2[3,2]
+; SSE2-NEXT:    movaps %xmm3, %xmm1
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: PR22412:
+; SSSE3:       # BB#0: # %entry
+; SSSE3-NEXT:    movsd {{.*#+}} xmm2 = xmm0[0],xmm2[1]
+; SSSE3-NEXT:    movapd %xmm2, %xmm0
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm3[3,2]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,0],xmm2[3,2]
+; SSSE3-NEXT:    movaps %xmm3, %xmm1
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: PR22412:
+; SSE41:       # BB#0: # %entry
+; SSE41-NEXT:    blendpd {{.*#+}} xmm0 = xmm0[0],xmm2[1]
+; SSE41-NEXT:    movapd %xmm0, %xmm1
+; SSE41-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,0],xmm3[3,2]
+; SSE41-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,0],xmm0[3,2]
+; SSE41-NEXT:    movaps %xmm1, %xmm0
+; SSE41-NEXT:    movaps %xmm3, %xmm1
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: PR22412:
+; AVX1:       # BB#0: # %entry
+; AVX1-NEXT:    vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
+; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX1-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[1,0],ymm1[3,2],ymm0[5,4],ymm1[7,6]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: PR22412:
+; AVX2:       # BB#0: # %entry
+; AVX2-NEXT:    vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm1 = [1,0,7,6,5,4,3,2]
+; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    retq
+entry:
+  %s1 = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 1, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+  %s2 = shufflevector <8 x float> %s1, <8 x float> undef, <8 x i32> <i32 1, i32 0, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2>
+  ret <8 x float> %s2
+}