From 46cba73063fd8b5a11b155615b2bd801b90b916f Mon Sep 17 00:00:00 2001
From: Craig Topper <craig.topper@intel.com>
Date: Mon, 22 Jul 2019 19:58:49 +0000
Subject: [PATCH] [X86] When using AND+PACKUS in lowerV16I8Shuffle, generate
 the build vector directly in v16i8 with the correct 0x00 or 0xFF elements
 rather than using another VT and bitcasting it.

The build_vector will become a constant pool load. By using the
desired type initially, it ensures we don't generate a bitcast
of the constant pool load which will need to be folded with
the load.

While experimenting with another patch, I noticed that when the
load type and the constant pool type don't match, then
SimplifyDemandedBits can't handle it. While we should probably
fix that, this was a simple way to fix the issue I saw.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@366732 91177308-0d34-0410-b5e6-96231b3b80d8
---
 lib/Target/X86/X86ISelLowering.cpp            |  9 ++++---
 test/CodeGen/X86/avg.ll                       |  2 +-
 test/CodeGen/X86/masked_store_trunc.ll        |  4 ++--
 test/CodeGen/X86/mmx-arith.ll                 |  4 ++--
 test/CodeGen/X86/oddshuffles.ll               |  4 ++--
 test/CodeGen/X86/oddsubvector.ll              |  2 +-
 test/CodeGen/X86/psubus.ll                    |  2 +-
 test/CodeGen/X86/sse2-intrinsics-canonical.ll |  2 +-
 test/CodeGen/X86/vector-reduce-and-bool.ll    |  4 ++--
 test/CodeGen/X86/vector-reduce-or-bool.ll     |  4 ++--
 test/CodeGen/X86/vector-reduce-xor-bool.ll    |  4 ++--
 test/CodeGen/X86/vector-shuffle-128-v16.ll    |  2 +-
 test/CodeGen/X86/vector-trunc-math-widen.ll   | 24 +++++++++----------
 test/CodeGen/X86/vector-trunc-math.ll         | 24 +++++++++----------
 test/CodeGen/X86/vector-trunc-widen.ll        |  6 ++---
 test/CodeGen/X86/vector-trunc.ll              |  6 ++---
 16 files changed, 51 insertions(+), 52 deletions(-)

diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index 41f19076a25..367590ac138 100644
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -14157,11 +14157,10 @@ static SDValue lowerV16I8Shuffle(const SDLoc &DL, ArrayRef<int> Mask,
     // First we need to zero all the dropped bytes.
     assert(NumEvenDrops <= 3 &&
            "No support for dropping even elements more than 3 times.");
-    // We use the mask type to pick which bytes are preserved based on how many
-    // elements are dropped.
-    MVT MaskVTs[] = { MVT::v8i16, MVT::v4i32, MVT::v2i64 };
-    SDValue ByteClearMask = DAG.getBitcast(
-        MVT::v16i8, DAG.getConstant(0xFF, DL, MaskVTs[NumEvenDrops - 1]));
+    SmallVector<SDValue, 16> ByteClearOps(16, DAG.getConstant(0, DL, MVT::i8));
+    for (unsigned i = 0; i != 16; i += 1 << NumEvenDrops)
+      ByteClearOps[i] = DAG.getConstant(0xFF, DL, MVT::i8);
+    SDValue ByteClearMask = DAG.getBuildVector(MVT::v16i8, DL, ByteClearOps);
     V1 = DAG.getNode(ISD::AND, DL, MVT::v16i8, V1, ByteClearMask);
     if (!IsSingleInput)
       V2 = DAG.getNode(ISD::AND, DL, MVT::v16i8, V2, ByteClearMask);
diff --git a/test/CodeGen/X86/avg.ll b/test/CodeGen/X86/avg.ll
index 86d8628c9ae..8a07b44bd50 100644
--- a/test/CodeGen/X86/avg.ll
+++ b/test/CodeGen/X86/avg.ll
@@ -226,7 +226,7 @@ define void @avg_v48i8(<48 x i8>* %a, <48 x i8>* %b) nounwind {
 ; SSE2-NEXT:    psubd %xmm5, %xmm0
 ; SSE2-NEXT:    psrld $1, %xmm3
 ; SSE2-NEXT:    psrld $1, %xmm8
-; SSE2-NEXT:    movdqa {{.*#+}} xmm7 = [255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm7 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
 ; SSE2-NEXT:    pand %xmm7, %xmm8
 ; SSE2-NEXT:    pand %xmm7, %xmm3
 ; SSE2-NEXT:    packuswb %xmm8, %xmm3
diff --git a/test/CodeGen/X86/masked_store_trunc.ll b/test/CodeGen/X86/masked_store_trunc.ll
index 8e73194bfa9..4111e83b599 100644
--- a/test/CodeGen/X86/masked_store_trunc.ll
+++ b/test/CodeGen/X86/masked_store_trunc.ll
@@ -4673,7 +4673,7 @@ define void @truncstore_v32i16_v32i8(<32 x i16> %x, <32 x i8>* %p, <32 x i8> %ma
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    pxor %xmm7, %xmm7
 ; SSE2-NEXT:    pcmpeqb %xmm4, %xmm7
-; SSE2-NEXT:    movdqa {{.*#+}} xmm6 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm6 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE2-NEXT:    pand %xmm6, %xmm1
 ; SSE2-NEXT:    pand %xmm6, %xmm0
 ; SSE2-NEXT:    packuswb %xmm1, %xmm0
@@ -6209,7 +6209,7 @@ define void @truncstore_v16i16_v16i8(<16 x i16> %x, <16 x i8>* %p, <16 x i8> %ma
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    pxor %xmm3, %xmm3
 ; SSE2-NEXT:    pcmpeqb %xmm2, %xmm3
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE2-NEXT:    pand %xmm4, %xmm1
 ; SSE2-NEXT:    pand %xmm4, %xmm0
 ; SSE2-NEXT:    packuswb %xmm1, %xmm0
diff --git a/test/CodeGen/X86/mmx-arith.ll b/test/CodeGen/X86/mmx-arith.ll
index 5df75efb1ea..a51fa2ac80e 100644
--- a/test/CodeGen/X86/mmx-arith.ll
+++ b/test/CodeGen/X86/mmx-arith.ll
@@ -42,7 +42,7 @@ define void @test0(x86_mmx* %A, x86_mmx* %B) {
 ; X32-NEXT:    movq {{.*#+}} xmm1 = mem[0],zero
 ; X32-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
 ; X32-NEXT:    pmullw %xmm0, %xmm1
-; X32-NEXT:    movdqa {{.*#+}} xmm0 = [255,255,255,255,255,255,255,255]
+; X32-NEXT:    movdqa {{.*#+}} xmm0 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; X32-NEXT:    movdqa %xmm1, %xmm2
 ; X32-NEXT:    pand %xmm0, %xmm2
 ; X32-NEXT:    packuswb %xmm2, %xmm2
@@ -100,7 +100,7 @@ define void @test0(x86_mmx* %A, x86_mmx* %B) {
 ; X64-NEXT:    movq {{.*#+}} xmm1 = mem[0],zero
 ; X64-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
 ; X64-NEXT:    pmullw %xmm0, %xmm1
-; X64-NEXT:    movdqa {{.*#+}} xmm0 = [255,255,255,255,255,255,255,255]
+; X64-NEXT:    movdqa {{.*#+}} xmm0 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; X64-NEXT:    movdqa %xmm1, %xmm2
 ; X64-NEXT:    pand %xmm0, %xmm2
 ; X64-NEXT:    packuswb %xmm2, %xmm2
diff --git a/test/CodeGen/X86/oddshuffles.ll b/test/CodeGen/X86/oddshuffles.ll
index 2c36385e714..2e66a22d711 100644
--- a/test/CodeGen/X86/oddshuffles.ll
+++ b/test/CodeGen/X86/oddshuffles.ll
@@ -259,7 +259,7 @@ define void @v7i8(<4 x i8> %a, <4 x i8> %b, <7 x i8>* %p) nounwind {
 ; SSE2-NEXT:    pand %xmm2, %xmm1
 ; SSE2-NEXT:    pandn %xmm0, %xmm2
 ; SSE2-NEXT:    por %xmm1, %xmm2
-; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE2-NEXT:    pand %xmm2, %xmm0
 ; SSE2-NEXT:    packuswb %xmm0, %xmm0
 ; SSE2-NEXT:    movdqa %xmm2, -{{[0-9]+}}(%rsp)
@@ -659,7 +659,7 @@ define void @v12i32(<8 x i32> %a, <8 x i32> %b, <12 x i32>* %p) nounwind {
 define void @pr29025(<4 x i8> %a, <4 x i8> %b, <4 x i8> %c, <12 x i8> *%p) nounwind {
 ; SSE2-LABEL: pr29025:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
 ; SSE2-NEXT:    pand %xmm3, %xmm1
 ; SSE2-NEXT:    pand %xmm3, %xmm0
 ; SSE2-NEXT:    packuswb %xmm1, %xmm0
diff --git a/test/CodeGen/X86/oddsubvector.ll b/test/CodeGen/X86/oddsubvector.ll
index 9bc6c0f380a..d9cc5232a83 100644
--- a/test/CodeGen/X86/oddsubvector.ll
+++ b/test/CodeGen/X86/oddsubvector.ll
@@ -17,7 +17,7 @@ define void @insert_v7i8_v2i16_2(<7 x i8> *%a0, <2 x i16> *%a1) nounwind {
 ; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
 ; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,3]
 ; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0,1,3]
-; SSE2-NEXT:    movaps {{.*#+}} xmm1 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    movaps {{.*#+}} xmm1 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE2-NEXT:    andps %xmm0, %xmm1
 ; SSE2-NEXT:    packuswb %xmm1, %xmm1
 ; SSE2-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
diff --git a/test/CodeGen/X86/psubus.ll b/test/CodeGen/X86/psubus.ll
index 1009b1ef906..351955fbb03 100644
--- a/test/CodeGen/X86/psubus.ll
+++ b/test/CodeGen/X86/psubus.ll
@@ -622,7 +622,7 @@ define <16 x i8> @test14(<16 x i8> %x, <16 x i32> %y) nounwind {
 ; SSE2-NEXT:    psubd %xmm5, %xmm4
 ; SSE2-NEXT:    por %xmm0, %xmm5
 ; SSE2-NEXT:    pcmpgtd %xmm9, %xmm5
-; SSE2-NEXT:    movdqa {{.*#+}} xmm9 = [255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm9 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
 ; SSE2-NEXT:    pand %xmm9, %xmm5
 ; SSE2-NEXT:    movdqa %xmm3, %xmm7
 ; SSE2-NEXT:    pxor %xmm0, %xmm7
diff --git a/test/CodeGen/X86/sse2-intrinsics-canonical.ll b/test/CodeGen/X86/sse2-intrinsics-canonical.ll
index 506fb9eb100..e8309e047b1 100644
--- a/test/CodeGen/X86/sse2-intrinsics-canonical.ll
+++ b/test/CodeGen/X86/sse2-intrinsics-canonical.ll
@@ -93,7 +93,7 @@ define <8 x i16> @test_x86_sse2_psubus_w(<8 x i16> %a0, <8 x i16> %a1) {
 define <8 x i8> @test_x86_sse2_paddus_b_64(<8 x i8> %a0, <8 x i8> %a1) {
 ; SSE-LABEL: test_x86_sse2_paddus_b_64:
 ; SSE:       ## %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    ## encoding: [0x66,0x0f,0x6f,0x15,A,A,A,A]
 ; SSE-NEXT:    ## fixup A - offset: 4, value: LCPI4_0, kind: FK_Data_4
 ; SSE-NEXT:    pand %xmm2, %xmm1 ## encoding: [0x66,0x0f,0xdb,0xca]
diff --git a/test/CodeGen/X86/vector-reduce-and-bool.ll b/test/CodeGen/X86/vector-reduce-and-bool.ll
index b8d57f6954d..91120dfcbd9 100644
--- a/test/CodeGen/X86/vector-reduce-and-bool.ll
+++ b/test/CodeGen/X86/vector-reduce-and-bool.ll
@@ -349,7 +349,7 @@ define i1 @trunc_v8i32_v8i1(<8 x i32>) {
 define i1 @trunc_v16i16_v16i1(<16 x i16>) {
 ; SSE2-LABEL: trunc_v16i16_v16i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE2-NEXT:    pand %xmm2, %xmm1
 ; SSE2-NEXT:    pand %xmm2, %xmm0
 ; SSE2-NEXT:    packuswb %xmm1, %xmm0
@@ -690,7 +690,7 @@ define i1 @trunc_v16i32_v16i1(<16 x i32>) {
 define i1 @trunc_v32i16_v32i1(<32 x i16>) {
 ; SSE2-LABEL: trunc_v32i16_v32i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE2-NEXT:    pand %xmm4, %xmm3
 ; SSE2-NEXT:    pand %xmm4, %xmm2
 ; SSE2-NEXT:    packuswb %xmm3, %xmm2
diff --git a/test/CodeGen/X86/vector-reduce-or-bool.ll b/test/CodeGen/X86/vector-reduce-or-bool.ll
index de5f52ffc92..7a411610384 100644
--- a/test/CodeGen/X86/vector-reduce-or-bool.ll
+++ b/test/CodeGen/X86/vector-reduce-or-bool.ll
@@ -340,7 +340,7 @@ define i1 @trunc_v8i32_v8i1(<8 x i32>) {
 define i1 @trunc_v16i16_v16i1(<16 x i16>) {
 ; SSE2-LABEL: trunc_v16i16_v16i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE2-NEXT:    pand %xmm2, %xmm1
 ; SSE2-NEXT:    pand %xmm2, %xmm0
 ; SSE2-NEXT:    packuswb %xmm1, %xmm0
@@ -681,7 +681,7 @@ define i1 @trunc_v16i32_v16i1(<16 x i32>) {
 define i1 @trunc_v32i16_v32i1(<32 x i16>) {
 ; SSE2-LABEL: trunc_v32i16_v32i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE2-NEXT:    pand %xmm4, %xmm3
 ; SSE2-NEXT:    pand %xmm4, %xmm2
 ; SSE2-NEXT:    packuswb %xmm3, %xmm2
diff --git a/test/CodeGen/X86/vector-reduce-xor-bool.ll b/test/CodeGen/X86/vector-reduce-xor-bool.ll
index eb0ba63fc12..059c0422b0a 100644
--- a/test/CodeGen/X86/vector-reduce-xor-bool.ll
+++ b/test/CodeGen/X86/vector-reduce-xor-bool.ll
@@ -355,7 +355,7 @@ define i1 @trunc_v8i32_v8i1(<8 x i32>) {
 define i1 @trunc_v16i16_v16i1(<16 x i16>) {
 ; SSE2-LABEL: trunc_v16i16_v16i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE2-NEXT:    pand %xmm2, %xmm1
 ; SSE2-NEXT:    pand %xmm2, %xmm0
 ; SSE2-NEXT:    packuswb %xmm1, %xmm0
@@ -767,7 +767,7 @@ define i1 @trunc_v16i32_v16i1(<16 x i32>) {
 define i1 @trunc_v32i16_v32i1(<32 x i16>) {
 ; SSE2-LABEL: trunc_v32i16_v32i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE2-NEXT:    pand %xmm4, %xmm3
 ; SSE2-NEXT:    pand %xmm4, %xmm2
 ; SSE2-NEXT:    packuswb %xmm3, %xmm2
diff --git a/test/CodeGen/X86/vector-shuffle-128-v16.ll b/test/CodeGen/X86/vector-shuffle-128-v16.ll
index 4df979adbdf..5b3ce4f9eab 100644
--- a/test/CodeGen/X86/vector-shuffle-128-v16.ll
+++ b/test/CodeGen/X86/vector-shuffle-128-v16.ll
@@ -1688,7 +1688,7 @@ define <16 x i8> @shuffle_v16i8_zz_zz_zz_zz_zz_zz_zz_zz_zz_zz_01_02_03_04_05_06(
 define <16 x i8> @PR12412(<16 x i8> %inval1, <16 x i8> %inval2) {
 ; SSE2-LABEL: PR12412:
 ; SSE2:       # %bb.0: # %entry
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE2-NEXT:    pand %xmm2, %xmm1
 ; SSE2-NEXT:    pand %xmm2, %xmm0
 ; SSE2-NEXT:    packuswb %xmm1, %xmm0
diff --git a/test/CodeGen/X86/vector-trunc-math-widen.ll b/test/CodeGen/X86/vector-trunc-math-widen.ll
index 5cba4a9342e..1c2813d35f0 100644
--- a/test/CodeGen/X86/vector-trunc-math-widen.ll
+++ b/test/CodeGen/X86/vector-trunc-math-widen.ll
@@ -389,7 +389,7 @@ define <16 x i8> @trunc_add_v16i16_v16i8(<16 x i16> %a0, <16 x i16> %a1) nounwin
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    paddw %xmm2, %xmm0
 ; SSE-NEXT:    paddw %xmm3, %xmm1
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -812,7 +812,7 @@ define <16 x i8> @trunc_add_const_v16i32_v16i8(<16 x i32> %a0) nounwind {
 define <16 x i8> @trunc_add_const_v16i16_v16i8(<16 x i16> %a0) nounwind {
 ; SSE-LABEL: trunc_add_const_v16i16_v16i8:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -1247,7 +1247,7 @@ define <16 x i8> @trunc_sub_v16i16_v16i8(<16 x i16> %a0, <16 x i16> %a1) nounwin
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    psubw %xmm2, %xmm0
 ; SSE-NEXT:    psubw %xmm3, %xmm1
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -1638,7 +1638,7 @@ define <16 x i8> @trunc_sub_const_v16i32_v16i8(<16 x i32> %a0) nounwind {
 define <16 x i8> @trunc_sub_const_v16i16_v16i8(<16 x i16> %a0) nounwind {
 ; SSE-LABEL: trunc_sub_const_v16i16_v16i8:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -2240,7 +2240,7 @@ define <16 x i8> @trunc_mul_v16i16_v16i8(<16 x i16> %a0, <16 x i16> %a1) nounwin
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    pmullw %xmm2, %xmm0
 ; SSE-NEXT:    pmullw %xmm3, %xmm1
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -2728,7 +2728,7 @@ define <16 x i8> @trunc_mul_const_v16i16_v16i8(<16 x i16> %a0) nounwind {
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    pmullw {{.*}}(%rip), %xmm0
 ; SSE-NEXT:    pmullw {{.*}}(%rip), %xmm1
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -3135,7 +3135,7 @@ define <16 x i8> @trunc_and_v16i32_v16i8(<16 x i32> %a0, <16 x i32> %a1) nounwin
 define <16 x i8> @trunc_and_v16i16_v16i8(<16 x i16> %a0, <16 x i16> %a1) nounwind {
 ; SSE-LABEL: trunc_and_v16i16_v16i8:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm4, %xmm3
 ; SSE-NEXT:    pand %xmm1, %xmm3
 ; SSE-NEXT:    pand %xmm4, %xmm2
@@ -3507,7 +3507,7 @@ define <16 x i8> @trunc_and_const_v16i32_v16i8(<16 x i32> %a0) nounwind {
 define <16 x i8> @trunc_and_const_v16i16_v16i8(<16 x i16> %a0) nounwind {
 ; SSE-LABEL: trunc_and_const_v16i16_v16i8:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -3914,7 +3914,7 @@ define <16 x i8> @trunc_xor_v16i16_v16i8(<16 x i16> %a0, <16 x i16> %a1) nounwin
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    pxor %xmm2, %xmm0
 ; SSE-NEXT:    pxor %xmm3, %xmm1
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -4284,7 +4284,7 @@ define <16 x i8> @trunc_xor_const_v16i32_v16i8(<16 x i32> %a0) nounwind {
 define <16 x i8> @trunc_xor_const_v16i16_v16i8(<16 x i16> %a0) nounwind {
 ; SSE-LABEL: trunc_xor_const_v16i16_v16i8:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -4691,7 +4691,7 @@ define <16 x i8> @trunc_or_v16i16_v16i8(<16 x i16> %a0, <16 x i16> %a1) nounwind
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    por %xmm2, %xmm0
 ; SSE-NEXT:    por %xmm3, %xmm1
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -5061,7 +5061,7 @@ define <16 x i8> @trunc_or_const_v16i32_v16i8(<16 x i32> %a0) nounwind {
 define <16 x i8> @trunc_or_const_v16i16_v16i8(<16 x i16> %a0) nounwind {
 ; SSE-LABEL: trunc_or_const_v16i16_v16i8:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
diff --git a/test/CodeGen/X86/vector-trunc-math.ll b/test/CodeGen/X86/vector-trunc-math.ll
index 3d2202e385a..b91b2592f46 100644
--- a/test/CodeGen/X86/vector-trunc-math.ll
+++ b/test/CodeGen/X86/vector-trunc-math.ll
@@ -389,7 +389,7 @@ define <16 x i8> @trunc_add_v16i16_v16i8(<16 x i16> %a0, <16 x i16> %a1) nounwin
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    paddw %xmm2, %xmm0
 ; SSE-NEXT:    paddw %xmm3, %xmm1
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -812,7 +812,7 @@ define <16 x i8> @trunc_add_const_v16i32_v16i8(<16 x i32> %a0) nounwind {
 define <16 x i8> @trunc_add_const_v16i16_v16i8(<16 x i16> %a0) nounwind {
 ; SSE-LABEL: trunc_add_const_v16i16_v16i8:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -1247,7 +1247,7 @@ define <16 x i8> @trunc_sub_v16i16_v16i8(<16 x i16> %a0, <16 x i16> %a1) nounwin
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    psubw %xmm2, %xmm0
 ; SSE-NEXT:    psubw %xmm3, %xmm1
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -1638,7 +1638,7 @@ define <16 x i8> @trunc_sub_const_v16i32_v16i8(<16 x i32> %a0) nounwind {
 define <16 x i8> @trunc_sub_const_v16i16_v16i8(<16 x i16> %a0) nounwind {
 ; SSE-LABEL: trunc_sub_const_v16i16_v16i8:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -2240,7 +2240,7 @@ define <16 x i8> @trunc_mul_v16i16_v16i8(<16 x i16> %a0, <16 x i16> %a1) nounwin
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    pmullw %xmm2, %xmm0
 ; SSE-NEXT:    pmullw %xmm3, %xmm1
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -2728,7 +2728,7 @@ define <16 x i8> @trunc_mul_const_v16i16_v16i8(<16 x i16> %a0) nounwind {
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    pmullw {{.*}}(%rip), %xmm0
 ; SSE-NEXT:    pmullw {{.*}}(%rip), %xmm1
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -3135,7 +3135,7 @@ define <16 x i8> @trunc_and_v16i32_v16i8(<16 x i32> %a0, <16 x i32> %a1) nounwin
 define <16 x i8> @trunc_and_v16i16_v16i8(<16 x i16> %a0, <16 x i16> %a1) nounwind {
 ; SSE-LABEL: trunc_and_v16i16_v16i8:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm4, %xmm3
 ; SSE-NEXT:    pand %xmm1, %xmm3
 ; SSE-NEXT:    pand %xmm4, %xmm2
@@ -3507,7 +3507,7 @@ define <16 x i8> @trunc_and_const_v16i32_v16i8(<16 x i32> %a0) nounwind {
 define <16 x i8> @trunc_and_const_v16i16_v16i8(<16 x i16> %a0) nounwind {
 ; SSE-LABEL: trunc_and_const_v16i16_v16i8:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -3914,7 +3914,7 @@ define <16 x i8> @trunc_xor_v16i16_v16i8(<16 x i16> %a0, <16 x i16> %a1) nounwin
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    pxor %xmm2, %xmm0
 ; SSE-NEXT:    pxor %xmm3, %xmm1
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -4284,7 +4284,7 @@ define <16 x i8> @trunc_xor_const_v16i32_v16i8(<16 x i32> %a0) nounwind {
 define <16 x i8> @trunc_xor_const_v16i16_v16i8(<16 x i16> %a0) nounwind {
 ; SSE-LABEL: trunc_xor_const_v16i16_v16i8:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -4691,7 +4691,7 @@ define <16 x i8> @trunc_or_v16i16_v16i8(<16 x i16> %a0, <16 x i16> %a1) nounwind
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    por %xmm2, %xmm0
 ; SSE-NEXT:    por %xmm3, %xmm1
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
@@ -5061,7 +5061,7 @@ define <16 x i8> @trunc_or_const_v16i32_v16i8(<16 x i32> %a0) nounwind {
 define <16 x i8> @trunc_or_const_v16i16_v16i8(<16 x i16> %a0) nounwind {
 ; SSE-LABEL: trunc_or_const_v16i16_v16i8:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE-NEXT:    pand %xmm2, %xmm1
 ; SSE-NEXT:    pand %xmm2, %xmm0
 ; SSE-NEXT:    packuswb %xmm1, %xmm0
diff --git a/test/CodeGen/X86/vector-trunc-widen.ll b/test/CodeGen/X86/vector-trunc-widen.ll
index 23a4a978a2d..42bf9a7c049 100644
--- a/test/CodeGen/X86/vector-trunc-widen.ll
+++ b/test/CodeGen/X86/vector-trunc-widen.ll
@@ -1042,7 +1042,7 @@ entry:
 define void @trunc16i16_16i8(<16 x i16> %a) {
 ; SSE2-LABEL: trunc16i16_16i8:
 ; SSE2:       # %bb.0: # %entry
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE2-NEXT:    pand %xmm2, %xmm1
 ; SSE2-NEXT:    pand %xmm2, %xmm0
 ; SSE2-NEXT:    packuswb %xmm1, %xmm0
@@ -1251,7 +1251,7 @@ entry:
 define void @trunc32i16_32i8(<32 x i16> %a) {
 ; SSE2-LABEL: trunc32i16_32i8:
 ; SSE2:       # %bb.0: # %entry
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE2-NEXT:    pand %xmm4, %xmm1
 ; SSE2-NEXT:    pand %xmm4, %xmm0
 ; SSE2-NEXT:    packuswb %xmm1, %xmm0
@@ -1722,7 +1722,7 @@ entry:
 define <16 x i8> @trunc2x8i16_16i8(<8 x i16> %a, <8 x i16> %b) {
 ; SSE2-LABEL: trunc2x8i16_16i8:
 ; SSE2:       # %bb.0: # %entry
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE2-NEXT:    pand %xmm2, %xmm1
 ; SSE2-NEXT:    pand %xmm2, %xmm0
 ; SSE2-NEXT:    packuswb %xmm1, %xmm0
diff --git a/test/CodeGen/X86/vector-trunc.ll b/test/CodeGen/X86/vector-trunc.ll
index 35b190b1132..d4874c18870 100644
--- a/test/CodeGen/X86/vector-trunc.ll
+++ b/test/CodeGen/X86/vector-trunc.ll
@@ -1052,7 +1052,7 @@ entry:
 define void @trunc16i16_16i8(<16 x i16> %a) {
 ; SSE2-LABEL: trunc16i16_16i8:
 ; SSE2:       # %bb.0: # %entry
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE2-NEXT:    pand %xmm2, %xmm1
 ; SSE2-NEXT:    pand %xmm2, %xmm0
 ; SSE2-NEXT:    packuswb %xmm1, %xmm0
@@ -1261,7 +1261,7 @@ entry:
 define void @trunc32i16_32i8(<32 x i16> %a) {
 ; SSE2-LABEL: trunc32i16_32i8:
 ; SSE2:       # %bb.0: # %entry
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE2-NEXT:    pand %xmm4, %xmm1
 ; SSE2-NEXT:    pand %xmm4, %xmm0
 ; SSE2-NEXT:    packuswb %xmm1, %xmm0
@@ -1728,7 +1728,7 @@ entry:
 define <16 x i8> @trunc2x8i16_16i8(<8 x i16> %a, <8 x i16> %b) {
 ; SSE2-LABEL: trunc2x8i16_16i8:
 ; SSE2:       # %bb.0: # %entry
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; SSE2-NEXT:    pand %xmm2, %xmm1
 ; SSE2-NEXT:    pand %xmm2, %xmm0
 ; SSE2-NEXT:    packuswb %xmm1, %xmm0
-- 
2.40.0