[clang] 1e5bfac - [clang][SystemZ][NFC] Autogenerate some test case

Ulrich Weigand via cfe-commits cfe-commits at lists.llvm.org
Wed Nov 13 07:15:00 PST 2024


Author: Ulrich Weigand
Date: 2024-11-13T16:14:37+01:00
New Revision: 1e5bfac933ea90ec4361446398551dd6b967c67f

URL: https://github.com/llvm/llvm-project/commit/1e5bfac933ea90ec4361446398551dd6b967c67f
DIFF: https://github.com/llvm/llvm-project/commit/1e5bfac933ea90ec4361446398551dd6b967c67f.diff

LOG: [clang][SystemZ][NFC] Autogenerate some test case

Autogenerate the clang/test/CodeGen/SystemZ/zvector{,2}.c
test cases to make it easier to update them in the future.

Added: 
    

Modified: 
    clang/test/CodeGen/SystemZ/zvector.c
    clang/test/CodeGen/SystemZ/zvector2.c

Removed: 
    


################################################################################
diff  --git a/clang/test/CodeGen/SystemZ/zvector.c b/clang/test/CodeGen/SystemZ/zvector.c
index 4b28e394b33b72..cbf6a9a1a1bf2a 100644
--- a/clang/test/CodeGen/SystemZ/zvector.c
+++ b/clang/test/CodeGen/SystemZ/zvector.c
@@ -1,3 +1,4 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
 // RUN: %clang_cc1 -triple s390x-linux-gnu -target-cpu z13 -fzvector \
 // RUN: -emit-llvm -o - -W -Wall -Werror \
 // RUN: %s | opt -S -passes=mem2reg | FileCheck %s
@@ -22,26 +23,29 @@ volatile vector double fd, fd2;
 
 volatile int cnt;
 
-// CHECK-LABEL: define{{.*}} void @test_assign() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   store volatile <16 x i8> [[TMP0]], ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   store volatile <16 x i8> [[TMP1]], ptr @uc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   store volatile <8 x i16> [[TMP2]], ptr @ss, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   store volatile <8 x i16> [[TMP3]], ptr @us, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   store volatile <4 x i32> [[TMP4]], ptr @si, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   store volatile <4 x i32> [[TMP5]], ptr @ui, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   store volatile <2 x i64> [[TMP6]], ptr @sl, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   store volatile <2 x i64> [[TMP7]], ptr @ul, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   store volatile <2 x double> [[TMP8]], ptr @fd, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_assign(
+// CHECK-SAME: ) #[[ATTR0:[0-9]+]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    store volatile <16 x i8> [[TMP0]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    store volatile <16 x i8> [[TMP1]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    store volatile <8 x i16> [[TMP2]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    store volatile <8 x i16> [[TMP3]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    store volatile <4 x i32> [[TMP4]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    store volatile <4 x i32> [[TMP5]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    store volatile <2 x i64> [[TMP6]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    store volatile <2 x i64> [[TMP7]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    store volatile <2 x double> [[TMP8]], ptr @fd, align 8
+// CHECK-NEXT:    ret void
+//
 void test_assign(void) {
 
   sc = sc2;
@@ -59,26 +63,29 @@ void test_assign(void) {
   fd = fd2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_pos() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   store volatile <16 x i8> [[TMP0]], ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   store volatile <16 x i8> [[TMP1]], ptr @uc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   store volatile <8 x i16> [[TMP2]], ptr @ss, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   store volatile <8 x i16> [[TMP3]], ptr @us, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   store volatile <4 x i32> [[TMP4]], ptr @si, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   store volatile <4 x i32> [[TMP5]], ptr @ui, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   store volatile <2 x i64> [[TMP6]], ptr @sl, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   store volatile <2 x i64> [[TMP7]], ptr @ul, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   store volatile <2 x double> [[TMP8]], ptr @fd, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_pos(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    store volatile <16 x i8> [[TMP0]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    store volatile <16 x i8> [[TMP1]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    store volatile <8 x i16> [[TMP2]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    store volatile <8 x i16> [[TMP3]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    store volatile <4 x i32> [[TMP4]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    store volatile <4 x i32> [[TMP5]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    store volatile <2 x i64> [[TMP6]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    store volatile <2 x i64> [[TMP7]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    store volatile <2 x double> [[TMP8]], ptr @fd, align 8
+// CHECK-NEXT:    ret void
+//
 void test_pos(void) {
 
   sc = +sc2;
@@ -96,23 +103,26 @@ void test_pos(void) {
   fd = +fd2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_neg() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[SUB:%.*]] = sub <16 x i8> zeroinitializer, [[TMP0]]
-// CHECK:   store volatile <16 x i8> [[SUB]], ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[SUB1:%.*]] = sub <8 x i16> zeroinitializer, [[TMP1]]
-// CHECK:   store volatile <8 x i16> [[SUB1]], ptr @ss, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[SUB2:%.*]] = sub <4 x i32> zeroinitializer, [[TMP2]]
-// CHECK:   store volatile <4 x i32> [[SUB2]], ptr @si, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[SUB3:%.*]] = sub <2 x i64> zeroinitializer, [[TMP3]]
-// CHECK:   store volatile <2 x i64> [[SUB3]], ptr @sl, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[SUB4:%.*]] = fneg <2 x double> [[TMP4]]
-// CHECK:   store volatile <2 x double> [[SUB4]], ptr @fd, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_neg(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[SUB:%.*]] = sub <16 x i8> zeroinitializer, [[TMP0]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SUB]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[SUB1:%.*]] = sub <8 x i16> zeroinitializer, [[TMP1]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SUB1]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[SUB2:%.*]] = sub <4 x i32> zeroinitializer, [[TMP2]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SUB2]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[SUB3:%.*]] = sub <2 x i64> zeroinitializer, [[TMP3]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SUB3]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[FNEG:%.*]] = fneg <2 x double> [[TMP4]]
+// CHECK-NEXT:    store volatile <2 x double> [[FNEG]], ptr @fd, align 8
+// CHECK-NEXT:    ret void
+//
 void test_neg(void) {
 
   sc = -sc2;
@@ -122,35 +132,38 @@ void test_neg(void) {
   fd = -fd2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_preinc() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[INC:%.*]] = add <16 x i8> [[TMP0]], splat (i8 1)
-// CHECK:   store volatile <16 x i8> [[INC]], ptr @sc2, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[INC1:%.*]] = add <16 x i8> [[TMP1]], splat (i8 1)
-// CHECK:   store volatile <16 x i8> [[INC1]], ptr @uc2, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[INC2:%.*]] = add <8 x i16> [[TMP2]], splat (i16 1)
-// CHECK:   store volatile <8 x i16> [[INC2]], ptr @ss2, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[INC3:%.*]] = add <8 x i16> [[TMP3]], splat (i16 1)
-// CHECK:   store volatile <8 x i16> [[INC3]], ptr @us2, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[INC4:%.*]] = add <4 x i32> [[TMP4]], splat (i32 1)
-// CHECK:   store volatile <4 x i32> [[INC4]], ptr @si2, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[INC5:%.*]] = add <4 x i32> [[TMP5]], splat (i32 1)
-// CHECK:   store volatile <4 x i32> [[INC5]], ptr @ui2, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[INC6:%.*]] = add <2 x i64> [[TMP6]], splat (i64 1)
-// CHECK:   store volatile <2 x i64> [[INC6]], ptr @sl2, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[INC7:%.*]] = add <2 x i64> [[TMP7]], splat (i64 1)
-// CHECK:   store volatile <2 x i64> [[INC7]], ptr @ul2, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[INC8:%.*]] = fadd <2 x double> [[TMP8]], splat (double 1.000000e+00)
-// CHECK:   store volatile <2 x double> [[INC8]], ptr @fd2, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_preinc(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[INC:%.*]] = add <16 x i8> [[TMP0]], splat (i8 1)
+// CHECK-NEXT:    store volatile <16 x i8> [[INC]], ptr @sc2, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[INC1:%.*]] = add <16 x i8> [[TMP1]], splat (i8 1)
+// CHECK-NEXT:    store volatile <16 x i8> [[INC1]], ptr @uc2, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[INC2:%.*]] = add <8 x i16> [[TMP2]], splat (i16 1)
+// CHECK-NEXT:    store volatile <8 x i16> [[INC2]], ptr @ss2, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[INC3:%.*]] = add <8 x i16> [[TMP3]], splat (i16 1)
+// CHECK-NEXT:    store volatile <8 x i16> [[INC3]], ptr @us2, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[INC4:%.*]] = add <4 x i32> [[TMP4]], splat (i32 1)
+// CHECK-NEXT:    store volatile <4 x i32> [[INC4]], ptr @si2, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[INC5:%.*]] = add <4 x i32> [[TMP5]], splat (i32 1)
+// CHECK-NEXT:    store volatile <4 x i32> [[INC5]], ptr @ui2, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[INC6:%.*]] = add <2 x i64> [[TMP6]], splat (i64 1)
+// CHECK-NEXT:    store volatile <2 x i64> [[INC6]], ptr @sl2, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[INC7:%.*]] = add <2 x i64> [[TMP7]], splat (i64 1)
+// CHECK-NEXT:    store volatile <2 x i64> [[INC7]], ptr @ul2, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[INC8:%.*]] = fadd <2 x double> [[TMP8]], splat (double 1.000000e+00)
+// CHECK-NEXT:    store volatile <2 x double> [[INC8]], ptr @fd2, align 8
+// CHECK-NEXT:    ret void
+//
 void test_preinc(void) {
 
   ++sc2;
@@ -168,35 +181,38 @@ void test_preinc(void) {
   ++fd2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_postinc() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[INC:%.*]] = add <16 x i8> [[TMP0]], splat (i8 1)
-// CHECK:   store volatile <16 x i8> [[INC]], ptr @sc2, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[INC1:%.*]] = add <16 x i8> [[TMP1]], splat (i8 1)
-// CHECK:   store volatile <16 x i8> [[INC1]], ptr @uc2, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[INC2:%.*]] = add <8 x i16> [[TMP2]], splat (i16 1)
-// CHECK:   store volatile <8 x i16> [[INC2]], ptr @ss2, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[INC3:%.*]] = add <8 x i16> [[TMP3]], splat (i16 1)
-// CHECK:   store volatile <8 x i16> [[INC3]], ptr @us2, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[INC4:%.*]] = add <4 x i32> [[TMP4]], splat (i32 1)
-// CHECK:   store volatile <4 x i32> [[INC4]], ptr @si2, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[INC5:%.*]] = add <4 x i32> [[TMP5]], splat (i32 1)
-// CHECK:   store volatile <4 x i32> [[INC5]], ptr @ui2, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[INC6:%.*]] = add <2 x i64> [[TMP6]], splat (i64 1)
-// CHECK:   store volatile <2 x i64> [[INC6]], ptr @sl2, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[INC7:%.*]] = add <2 x i64> [[TMP7]], splat (i64 1)
-// CHECK:   store volatile <2 x i64> [[INC7]], ptr @ul2, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[INC8:%.*]] = fadd <2 x double> [[TMP8]], splat (double 1.000000e+00)
-// CHECK:   store volatile <2 x double> [[INC8]], ptr @fd2, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_postinc(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[INC:%.*]] = add <16 x i8> [[TMP0]], splat (i8 1)
+// CHECK-NEXT:    store volatile <16 x i8> [[INC]], ptr @sc2, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[INC1:%.*]] = add <16 x i8> [[TMP1]], splat (i8 1)
+// CHECK-NEXT:    store volatile <16 x i8> [[INC1]], ptr @uc2, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[INC2:%.*]] = add <8 x i16> [[TMP2]], splat (i16 1)
+// CHECK-NEXT:    store volatile <8 x i16> [[INC2]], ptr @ss2, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[INC3:%.*]] = add <8 x i16> [[TMP3]], splat (i16 1)
+// CHECK-NEXT:    store volatile <8 x i16> [[INC3]], ptr @us2, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[INC4:%.*]] = add <4 x i32> [[TMP4]], splat (i32 1)
+// CHECK-NEXT:    store volatile <4 x i32> [[INC4]], ptr @si2, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[INC5:%.*]] = add <4 x i32> [[TMP5]], splat (i32 1)
+// CHECK-NEXT:    store volatile <4 x i32> [[INC5]], ptr @ui2, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[INC6:%.*]] = add <2 x i64> [[TMP6]], splat (i64 1)
+// CHECK-NEXT:    store volatile <2 x i64> [[INC6]], ptr @sl2, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[INC7:%.*]] = add <2 x i64> [[TMP7]], splat (i64 1)
+// CHECK-NEXT:    store volatile <2 x i64> [[INC7]], ptr @ul2, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[INC8:%.*]] = fadd <2 x double> [[TMP8]], splat (double 1.000000e+00)
+// CHECK-NEXT:    store volatile <2 x double> [[INC8]], ptr @fd2, align 8
+// CHECK-NEXT:    ret void
+//
 void test_postinc(void) {
 
   sc2++;
@@ -214,35 +230,38 @@ void test_postinc(void) {
   fd2++;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_predec() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[DEC:%.*]] = add <16 x i8> [[TMP0]], splat (i8 -1)
-// CHECK:   store volatile <16 x i8> [[DEC]], ptr @sc2, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[DEC1:%.*]] = add <16 x i8> [[TMP1]], splat (i8 -1)
-// CHECK:   store volatile <16 x i8> [[DEC1]], ptr @uc2, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[DEC2:%.*]] = add <8 x i16> [[TMP2]], splat (i16 -1)
-// CHECK:   store volatile <8 x i16> [[DEC2]], ptr @ss2, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[DEC3:%.*]] = add <8 x i16> [[TMP3]], splat (i16 -1)
-// CHECK:   store volatile <8 x i16> [[DEC3]], ptr @us2, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[DEC4:%.*]] = add <4 x i32> [[TMP4]], splat (i32 -1)
-// CHECK:   store volatile <4 x i32> [[DEC4]], ptr @si2, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[DEC5:%.*]] = add <4 x i32> [[TMP5]], splat (i32 -1)
-// CHECK:   store volatile <4 x i32> [[DEC5]], ptr @ui2, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[DEC6:%.*]] = add <2 x i64> [[TMP6]], splat (i64 -1)
-// CHECK:   store volatile <2 x i64> [[DEC6]], ptr @sl2, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[DEC7:%.*]] = add <2 x i64> [[TMP7]], splat (i64 -1)
-// CHECK:   store volatile <2 x i64> [[DEC7]], ptr @ul2, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[DEC8:%.*]] = fadd <2 x double> [[TMP8]], splat (double -1.000000e+00)
-// CHECK:   store volatile <2 x double> [[DEC8]], ptr @fd2, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_predec(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[DEC:%.*]] = add <16 x i8> [[TMP0]], splat (i8 -1)
+// CHECK-NEXT:    store volatile <16 x i8> [[DEC]], ptr @sc2, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[DEC1:%.*]] = add <16 x i8> [[TMP1]], splat (i8 -1)
+// CHECK-NEXT:    store volatile <16 x i8> [[DEC1]], ptr @uc2, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[DEC2:%.*]] = add <8 x i16> [[TMP2]], splat (i16 -1)
+// CHECK-NEXT:    store volatile <8 x i16> [[DEC2]], ptr @ss2, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[DEC3:%.*]] = add <8 x i16> [[TMP3]], splat (i16 -1)
+// CHECK-NEXT:    store volatile <8 x i16> [[DEC3]], ptr @us2, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[DEC4:%.*]] = add <4 x i32> [[TMP4]], splat (i32 -1)
+// CHECK-NEXT:    store volatile <4 x i32> [[DEC4]], ptr @si2, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[DEC5:%.*]] = add <4 x i32> [[TMP5]], splat (i32 -1)
+// CHECK-NEXT:    store volatile <4 x i32> [[DEC5]], ptr @ui2, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[DEC6:%.*]] = add <2 x i64> [[TMP6]], splat (i64 -1)
+// CHECK-NEXT:    store volatile <2 x i64> [[DEC6]], ptr @sl2, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[DEC7:%.*]] = add <2 x i64> [[TMP7]], splat (i64 -1)
+// CHECK-NEXT:    store volatile <2 x i64> [[DEC7]], ptr @ul2, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[DEC8:%.*]] = fadd <2 x double> [[TMP8]], splat (double -1.000000e+00)
+// CHECK-NEXT:    store volatile <2 x double> [[DEC8]], ptr @fd2, align 8
+// CHECK-NEXT:    ret void
+//
 void test_predec(void) {
 
   --sc2;
@@ -260,35 +279,38 @@ void test_predec(void) {
   --fd2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_postdec() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[DEC:%.*]] = add <16 x i8> [[TMP0]], splat (i8 -1)
-// CHECK:   store volatile <16 x i8> [[DEC]], ptr @sc2, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[DEC1:%.*]] = add <16 x i8> [[TMP1]], splat (i8 -1)
-// CHECK:   store volatile <16 x i8> [[DEC1]], ptr @uc2, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[DEC2:%.*]] = add <8 x i16> [[TMP2]], splat (i16 -1)
-// CHECK:   store volatile <8 x i16> [[DEC2]], ptr @ss2, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[DEC3:%.*]] = add <8 x i16> [[TMP3]], splat (i16 -1)
-// CHECK:   store volatile <8 x i16> [[DEC3]], ptr @us2, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[DEC4:%.*]] = add <4 x i32> [[TMP4]], splat (i32 -1)
-// CHECK:   store volatile <4 x i32> [[DEC4]], ptr @si2, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[DEC5:%.*]] = add <4 x i32> [[TMP5]], splat (i32 -1)
-// CHECK:   store volatile <4 x i32> [[DEC5]], ptr @ui2, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[DEC6:%.*]] = add <2 x i64> [[TMP6]], splat (i64 -1)
-// CHECK:   store volatile <2 x i64> [[DEC6]], ptr @sl2, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[DEC7:%.*]] = add <2 x i64> [[TMP7]], splat (i64 -1)
-// CHECK:   store volatile <2 x i64> [[DEC7]], ptr @ul2, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[DEC8:%.*]] = fadd <2 x double> [[TMP8]], splat (double -1.000000e+00)
-// CHECK:   store volatile <2 x double> [[DEC8]], ptr @fd2, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_postdec(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[DEC:%.*]] = add <16 x i8> [[TMP0]], splat (i8 -1)
+// CHECK-NEXT:    store volatile <16 x i8> [[DEC]], ptr @sc2, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[DEC1:%.*]] = add <16 x i8> [[TMP1]], splat (i8 -1)
+// CHECK-NEXT:    store volatile <16 x i8> [[DEC1]], ptr @uc2, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[DEC2:%.*]] = add <8 x i16> [[TMP2]], splat (i16 -1)
+// CHECK-NEXT:    store volatile <8 x i16> [[DEC2]], ptr @ss2, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[DEC3:%.*]] = add <8 x i16> [[TMP3]], splat (i16 -1)
+// CHECK-NEXT:    store volatile <8 x i16> [[DEC3]], ptr @us2, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[DEC4:%.*]] = add <4 x i32> [[TMP4]], splat (i32 -1)
+// CHECK-NEXT:    store volatile <4 x i32> [[DEC4]], ptr @si2, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[DEC5:%.*]] = add <4 x i32> [[TMP5]], splat (i32 -1)
+// CHECK-NEXT:    store volatile <4 x i32> [[DEC5]], ptr @ui2, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[DEC6:%.*]] = add <2 x i64> [[TMP6]], splat (i64 -1)
+// CHECK-NEXT:    store volatile <2 x i64> [[DEC6]], ptr @sl2, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[DEC7:%.*]] = add <2 x i64> [[TMP7]], splat (i64 -1)
+// CHECK-NEXT:    store volatile <2 x i64> [[DEC7]], ptr @ul2, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[DEC8:%.*]] = fadd <2 x double> [[TMP8]], splat (double -1.000000e+00)
+// CHECK-NEXT:    store volatile <2 x double> [[DEC8]], ptr @fd2, align 8
+// CHECK-NEXT:    ret void
+//
 void test_postdec(void) {
 
   sc2--;
@@ -306,108 +328,111 @@ void test_postdec(void) {
   fd2--;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_add() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[ADD:%.*]] = add <16 x i8> [[TMP0]], [[TMP1]]
-// CHECK:   store volatile <16 x i8> [[ADD]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[ADD1:%.*]] = add <16 x i8> [[TMP2]], [[TMP3]]
-// CHECK:   store volatile <16 x i8> [[ADD1]], ptr @sc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[ADD2:%.*]] = add <16 x i8> [[TMP4]], [[TMP5]]
-// CHECK:   store volatile <16 x i8> [[ADD2]], ptr @sc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[ADD3:%.*]] = add <16 x i8> [[TMP6]], [[TMP7]]
-// CHECK:   store volatile <16 x i8> [[ADD3]], ptr @uc, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[ADD4:%.*]] = add <16 x i8> [[TMP8]], [[TMP9]]
-// CHECK:   store volatile <16 x i8> [[ADD4]], ptr @uc, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[ADD5:%.*]] = add <16 x i8> [[TMP10]], [[TMP11]]
-// CHECK:   store volatile <16 x i8> [[ADD5]], ptr @uc, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[ADD6:%.*]] = add <8 x i16> [[TMP12]], [[TMP13]]
-// CHECK:   store volatile <8 x i16> [[ADD6]], ptr @ss, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[ADD7:%.*]] = add <8 x i16> [[TMP14]], [[TMP15]]
-// CHECK:   store volatile <8 x i16> [[ADD7]], ptr @ss, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[ADD8:%.*]] = add <8 x i16> [[TMP16]], [[TMP17]]
-// CHECK:   store volatile <8 x i16> [[ADD8]], ptr @ss, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP19:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[ADD9:%.*]] = add <8 x i16> [[TMP18]], [[TMP19]]
-// CHECK:   store volatile <8 x i16> [[ADD9]], ptr @us, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[ADD10:%.*]] = add <8 x i16> [[TMP20]], [[TMP21]]
-// CHECK:   store volatile <8 x i16> [[ADD10]], ptr @us, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[ADD11:%.*]] = add <8 x i16> [[TMP22]], [[TMP23]]
-// CHECK:   store volatile <8 x i16> [[ADD11]], ptr @us, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[ADD12:%.*]] = add <4 x i32> [[TMP24]], [[TMP25]]
-// CHECK:   store volatile <4 x i32> [[ADD12]], ptr @si, align 8
-// CHECK:   [[TMP26:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP27:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[ADD13:%.*]] = add <4 x i32> [[TMP26]], [[TMP27]]
-// CHECK:   store volatile <4 x i32> [[ADD13]], ptr @si, align 8
-// CHECK:   [[TMP28:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[ADD14:%.*]] = add <4 x i32> [[TMP28]], [[TMP29]]
-// CHECK:   store volatile <4 x i32> [[ADD14]], ptr @si, align 8
-// CHECK:   [[TMP30:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP31:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[ADD15:%.*]] = add <4 x i32> [[TMP30]], [[TMP31]]
-// CHECK:   store volatile <4 x i32> [[ADD15]], ptr @ui, align 8
-// CHECK:   [[TMP32:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP33:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[ADD16:%.*]] = add <4 x i32> [[TMP32]], [[TMP33]]
-// CHECK:   store volatile <4 x i32> [[ADD16]], ptr @ui, align 8
-// CHECK:   [[TMP34:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP35:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[ADD17:%.*]] = add <4 x i32> [[TMP34]], [[TMP35]]
-// CHECK:   store volatile <4 x i32> [[ADD17]], ptr @ui, align 8
-// CHECK:   [[TMP36:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP37:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[ADD18:%.*]] = add <2 x i64> [[TMP36]], [[TMP37]]
-// CHECK:   store volatile <2 x i64> [[ADD18]], ptr @sl, align 8
-// CHECK:   [[TMP38:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP39:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[ADD19:%.*]] = add <2 x i64> [[TMP38]], [[TMP39]]
-// CHECK:   store volatile <2 x i64> [[ADD19]], ptr @sl, align 8
-// CHECK:   [[TMP40:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP41:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[ADD20:%.*]] = add <2 x i64> [[TMP40]], [[TMP41]]
-// CHECK:   store volatile <2 x i64> [[ADD20]], ptr @sl, align 8
-// CHECK:   [[TMP42:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP43:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[ADD21:%.*]] = add <2 x i64> [[TMP42]], [[TMP43]]
-// CHECK:   store volatile <2 x i64> [[ADD21]], ptr @ul, align 8
-// CHECK:   [[TMP44:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP45:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[ADD22:%.*]] = add <2 x i64> [[TMP44]], [[TMP45]]
-// CHECK:   store volatile <2 x i64> [[ADD22]], ptr @ul, align 8
-// CHECK:   [[TMP46:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP47:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[ADD23:%.*]] = add <2 x i64> [[TMP46]], [[TMP47]]
-// CHECK:   store volatile <2 x i64> [[ADD23]], ptr @ul, align 8
-// CHECK:   [[TMP48:%.*]] = load volatile <2 x double>, ptr @fd, align 8
-// CHECK:   [[TMP49:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[ADD24:%.*]] = fadd <2 x double> [[TMP48]], [[TMP49]]
-// CHECK:   store volatile <2 x double> [[ADD24]], ptr @fd, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_add(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[ADD:%.*]] = add <16 x i8> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    store volatile <16 x i8> [[ADD]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[ADD1:%.*]] = add <16 x i8> [[TMP2]], [[TMP3]]
+// CHECK-NEXT:    store volatile <16 x i8> [[ADD1]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[ADD2:%.*]] = add <16 x i8> [[TMP4]], [[TMP5]]
+// CHECK-NEXT:    store volatile <16 x i8> [[ADD2]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[ADD3:%.*]] = add <16 x i8> [[TMP6]], [[TMP7]]
+// CHECK-NEXT:    store volatile <16 x i8> [[ADD3]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[ADD4:%.*]] = add <16 x i8> [[TMP8]], [[TMP9]]
+// CHECK-NEXT:    store volatile <16 x i8> [[ADD4]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[ADD5:%.*]] = add <16 x i8> [[TMP10]], [[TMP11]]
+// CHECK-NEXT:    store volatile <16 x i8> [[ADD5]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[ADD6:%.*]] = add <8 x i16> [[TMP12]], [[TMP13]]
+// CHECK-NEXT:    store volatile <8 x i16> [[ADD6]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[ADD7:%.*]] = add <8 x i16> [[TMP14]], [[TMP15]]
+// CHECK-NEXT:    store volatile <8 x i16> [[ADD7]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[ADD8:%.*]] = add <8 x i16> [[TMP16]], [[TMP17]]
+// CHECK-NEXT:    store volatile <8 x i16> [[ADD8]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[ADD9:%.*]] = add <8 x i16> [[TMP18]], [[TMP19]]
+// CHECK-NEXT:    store volatile <8 x i16> [[ADD9]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[ADD10:%.*]] = add <8 x i16> [[TMP20]], [[TMP21]]
+// CHECK-NEXT:    store volatile <8 x i16> [[ADD10]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[ADD11:%.*]] = add <8 x i16> [[TMP22]], [[TMP23]]
+// CHECK-NEXT:    store volatile <8 x i16> [[ADD11]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[ADD12:%.*]] = add <4 x i32> [[TMP24]], [[TMP25]]
+// CHECK-NEXT:    store volatile <4 x i32> [[ADD12]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP26:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP27:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[ADD13:%.*]] = add <4 x i32> [[TMP26]], [[TMP27]]
+// CHECK-NEXT:    store volatile <4 x i32> [[ADD13]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP28:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[ADD14:%.*]] = add <4 x i32> [[TMP28]], [[TMP29]]
+// CHECK-NEXT:    store volatile <4 x i32> [[ADD14]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP30:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP31:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[ADD15:%.*]] = add <4 x i32> [[TMP30]], [[TMP31]]
+// CHECK-NEXT:    store volatile <4 x i32> [[ADD15]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP32:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP33:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[ADD16:%.*]] = add <4 x i32> [[TMP32]], [[TMP33]]
+// CHECK-NEXT:    store volatile <4 x i32> [[ADD16]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP34:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP35:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[ADD17:%.*]] = add <4 x i32> [[TMP34]], [[TMP35]]
+// CHECK-NEXT:    store volatile <4 x i32> [[ADD17]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP36:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP37:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[ADD18:%.*]] = add <2 x i64> [[TMP36]], [[TMP37]]
+// CHECK-NEXT:    store volatile <2 x i64> [[ADD18]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP38:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP39:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[ADD19:%.*]] = add <2 x i64> [[TMP38]], [[TMP39]]
+// CHECK-NEXT:    store volatile <2 x i64> [[ADD19]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP40:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP41:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[ADD20:%.*]] = add <2 x i64> [[TMP40]], [[TMP41]]
+// CHECK-NEXT:    store volatile <2 x i64> [[ADD20]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP42:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP43:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[ADD21:%.*]] = add <2 x i64> [[TMP42]], [[TMP43]]
+// CHECK-NEXT:    store volatile <2 x i64> [[ADD21]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP44:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP45:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[ADD22:%.*]] = add <2 x i64> [[TMP44]], [[TMP45]]
+// CHECK-NEXT:    store volatile <2 x i64> [[ADD22]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP46:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP47:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[ADD23:%.*]] = add <2 x i64> [[TMP46]], [[TMP47]]
+// CHECK-NEXT:    store volatile <2 x i64> [[ADD23]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP48:%.*]] = load volatile <2 x double>, ptr @fd, align 8
+// CHECK-NEXT:    [[TMP49:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[ADD24:%.*]] = fadd <2 x double> [[TMP48]], [[TMP49]]
+// CHECK-NEXT:    store volatile <2 x double> [[ADD24]], ptr @fd, align 8
+// CHECK-NEXT:    ret void
+//
 void test_add(void) {
 
   sc = sc + sc2;
@@ -441,76 +466,79 @@ void test_add(void) {
   fd = fd + fd2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_add_assign() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[ADD:%.*]] = add <16 x i8> [[TMP1]], [[TMP0]]
-// CHECK:   store volatile <16 x i8> [[ADD]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[ADD1:%.*]] = add <16 x i8> [[TMP3]], [[TMP2]]
-// CHECK:   store volatile <16 x i8> [[ADD1]], ptr @sc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[ADD2:%.*]] = add <16 x i8> [[TMP5]], [[TMP4]]
-// CHECK:   store volatile <16 x i8> [[ADD2]], ptr @uc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[ADD3:%.*]] = add <16 x i8> [[TMP7]], [[TMP6]]
-// CHECK:   store volatile <16 x i8> [[ADD3]], ptr @uc, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[ADD4:%.*]] = add <8 x i16> [[TMP9]], [[TMP8]]
-// CHECK:   store volatile <8 x i16> [[ADD4]], ptr @ss, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[ADD5:%.*]] = add <8 x i16> [[TMP11]], [[TMP10]]
-// CHECK:   store volatile <8 x i16> [[ADD5]], ptr @ss, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[ADD6:%.*]] = add <8 x i16> [[TMP13]], [[TMP12]]
-// CHECK:   store volatile <8 x i16> [[ADD6]], ptr @us, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[ADD7:%.*]] = add <8 x i16> [[TMP15]], [[TMP14]]
-// CHECK:   store volatile <8 x i16> [[ADD7]], ptr @us, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[ADD8:%.*]] = add <4 x i32> [[TMP17]], [[TMP16]]
-// CHECK:   store volatile <4 x i32> [[ADD8]], ptr @si, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[TMP19:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[ADD9:%.*]] = add <4 x i32> [[TMP19]], [[TMP18]]
-// CHECK:   store volatile <4 x i32> [[ADD9]], ptr @si, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[ADD10:%.*]] = add <4 x i32> [[TMP21]], [[TMP20]]
-// CHECK:   store volatile <4 x i32> [[ADD10]], ptr @ui, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[ADD11:%.*]] = add <4 x i32> [[TMP23]], [[TMP22]]
-// CHECK:   store volatile <4 x i32> [[ADD11]], ptr @ui, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[ADD12:%.*]] = add <2 x i64> [[TMP25]], [[TMP24]]
-// CHECK:   store volatile <2 x i64> [[ADD12]], ptr @sl, align 8
-// CHECK:   [[TMP26:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[TMP27:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[ADD13:%.*]] = add <2 x i64> [[TMP27]], [[TMP26]]
-// CHECK:   store volatile <2 x i64> [[ADD13]], ptr @sl, align 8
-// CHECK:   [[TMP28:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[TMP29:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[ADD14:%.*]] = add <2 x i64> [[TMP29]], [[TMP28]]
-// CHECK:   store volatile <2 x i64> [[ADD14]], ptr @ul, align 8
-// CHECK:   [[TMP30:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[TMP31:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[ADD15:%.*]] = add <2 x i64> [[TMP31]], [[TMP30]]
-// CHECK:   store volatile <2 x i64> [[ADD15]], ptr @ul, align 8
-// CHECK:   [[TMP32:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[TMP33:%.*]] = load volatile <2 x double>, ptr @fd, align 8
-// CHECK:   [[ADD16:%.*]] = fadd <2 x double> [[TMP33]], [[TMP32]]
-// CHECK:   store volatile <2 x double> [[ADD16]], ptr @fd, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_add_assign(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[ADD:%.*]] = add <16 x i8> [[TMP1]], [[TMP0]]
+// CHECK-NEXT:    store volatile <16 x i8> [[ADD]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[ADD1:%.*]] = add <16 x i8> [[TMP3]], [[TMP2]]
+// CHECK-NEXT:    store volatile <16 x i8> [[ADD1]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[ADD2:%.*]] = add <16 x i8> [[TMP5]], [[TMP4]]
+// CHECK-NEXT:    store volatile <16 x i8> [[ADD2]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[ADD3:%.*]] = add <16 x i8> [[TMP7]], [[TMP6]]
+// CHECK-NEXT:    store volatile <16 x i8> [[ADD3]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[ADD4:%.*]] = add <8 x i16> [[TMP9]], [[TMP8]]
+// CHECK-NEXT:    store volatile <8 x i16> [[ADD4]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[ADD5:%.*]] = add <8 x i16> [[TMP11]], [[TMP10]]
+// CHECK-NEXT:    store volatile <8 x i16> [[ADD5]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[ADD6:%.*]] = add <8 x i16> [[TMP13]], [[TMP12]]
+// CHECK-NEXT:    store volatile <8 x i16> [[ADD6]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[ADD7:%.*]] = add <8 x i16> [[TMP15]], [[TMP14]]
+// CHECK-NEXT:    store volatile <8 x i16> [[ADD7]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[ADD8:%.*]] = add <4 x i32> [[TMP17]], [[TMP16]]
+// CHECK-NEXT:    store volatile <4 x i32> [[ADD8]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[ADD9:%.*]] = add <4 x i32> [[TMP19]], [[TMP18]]
+// CHECK-NEXT:    store volatile <4 x i32> [[ADD9]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[ADD10:%.*]] = add <4 x i32> [[TMP21]], [[TMP20]]
+// CHECK-NEXT:    store volatile <4 x i32> [[ADD10]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[ADD11:%.*]] = add <4 x i32> [[TMP23]], [[TMP22]]
+// CHECK-NEXT:    store volatile <4 x i32> [[ADD11]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[ADD12:%.*]] = add <2 x i64> [[TMP25]], [[TMP24]]
+// CHECK-NEXT:    store volatile <2 x i64> [[ADD12]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP26:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[TMP27:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[ADD13:%.*]] = add <2 x i64> [[TMP27]], [[TMP26]]
+// CHECK-NEXT:    store volatile <2 x i64> [[ADD13]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP28:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[TMP29:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[ADD14:%.*]] = add <2 x i64> [[TMP29]], [[TMP28]]
+// CHECK-NEXT:    store volatile <2 x i64> [[ADD14]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP30:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[TMP31:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[ADD15:%.*]] = add <2 x i64> [[TMP31]], [[TMP30]]
+// CHECK-NEXT:    store volatile <2 x i64> [[ADD15]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP32:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[TMP33:%.*]] = load volatile <2 x double>, ptr @fd, align 8
+// CHECK-NEXT:    [[ADD16:%.*]] = fadd <2 x double> [[TMP33]], [[TMP32]]
+// CHECK-NEXT:    store volatile <2 x double> [[ADD16]], ptr @fd, align 8
+// CHECK-NEXT:    ret void
+//
 void test_add_assign(void) {
 
   sc += sc2;
@@ -536,108 +564,111 @@ void test_add_assign(void) {
   fd += fd2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_sub() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[SUB:%.*]] = sub <16 x i8> [[TMP0]], [[TMP1]]
-// CHECK:   store volatile <16 x i8> [[SUB]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[SUB1:%.*]] = sub <16 x i8> [[TMP2]], [[TMP3]]
-// CHECK:   store volatile <16 x i8> [[SUB1]], ptr @sc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[SUB2:%.*]] = sub <16 x i8> [[TMP4]], [[TMP5]]
-// CHECK:   store volatile <16 x i8> [[SUB2]], ptr @sc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[SUB3:%.*]] = sub <16 x i8> [[TMP6]], [[TMP7]]
-// CHECK:   store volatile <16 x i8> [[SUB3]], ptr @uc, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[SUB4:%.*]] = sub <16 x i8> [[TMP8]], [[TMP9]]
-// CHECK:   store volatile <16 x i8> [[SUB4]], ptr @uc, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[SUB5:%.*]] = sub <16 x i8> [[TMP10]], [[TMP11]]
-// CHECK:   store volatile <16 x i8> [[SUB5]], ptr @uc, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[SUB6:%.*]] = sub <8 x i16> [[TMP12]], [[TMP13]]
-// CHECK:   store volatile <8 x i16> [[SUB6]], ptr @ss, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[SUB7:%.*]] = sub <8 x i16> [[TMP14]], [[TMP15]]
-// CHECK:   store volatile <8 x i16> [[SUB7]], ptr @ss, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[SUB8:%.*]] = sub <8 x i16> [[TMP16]], [[TMP17]]
-// CHECK:   store volatile <8 x i16> [[SUB8]], ptr @ss, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP19:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[SUB9:%.*]] = sub <8 x i16> [[TMP18]], [[TMP19]]
-// CHECK:   store volatile <8 x i16> [[SUB9]], ptr @us, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[SUB10:%.*]] = sub <8 x i16> [[TMP20]], [[TMP21]]
-// CHECK:   store volatile <8 x i16> [[SUB10]], ptr @us, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[SUB11:%.*]] = sub <8 x i16> [[TMP22]], [[TMP23]]
-// CHECK:   store volatile <8 x i16> [[SUB11]], ptr @us, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[SUB12:%.*]] = sub <4 x i32> [[TMP24]], [[TMP25]]
-// CHECK:   store volatile <4 x i32> [[SUB12]], ptr @si, align 8
-// CHECK:   [[TMP26:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP27:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[SUB13:%.*]] = sub <4 x i32> [[TMP26]], [[TMP27]]
-// CHECK:   store volatile <4 x i32> [[SUB13]], ptr @si, align 8
-// CHECK:   [[TMP28:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[SUB14:%.*]] = sub <4 x i32> [[TMP28]], [[TMP29]]
-// CHECK:   store volatile <4 x i32> [[SUB14]], ptr @si, align 8
-// CHECK:   [[TMP30:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP31:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[SUB15:%.*]] = sub <4 x i32> [[TMP30]], [[TMP31]]
-// CHECK:   store volatile <4 x i32> [[SUB15]], ptr @ui, align 8
-// CHECK:   [[TMP32:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP33:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[SUB16:%.*]] = sub <4 x i32> [[TMP32]], [[TMP33]]
-// CHECK:   store volatile <4 x i32> [[SUB16]], ptr @ui, align 8
-// CHECK:   [[TMP34:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP35:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[SUB17:%.*]] = sub <4 x i32> [[TMP34]], [[TMP35]]
-// CHECK:   store volatile <4 x i32> [[SUB17]], ptr @ui, align 8
-// CHECK:   [[TMP36:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP37:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[SUB18:%.*]] = sub <2 x i64> [[TMP36]], [[TMP37]]
-// CHECK:   store volatile <2 x i64> [[SUB18]], ptr @sl, align 8
-// CHECK:   [[TMP38:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP39:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[SUB19:%.*]] = sub <2 x i64> [[TMP38]], [[TMP39]]
-// CHECK:   store volatile <2 x i64> [[SUB19]], ptr @sl, align 8
-// CHECK:   [[TMP40:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP41:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[SUB20:%.*]] = sub <2 x i64> [[TMP40]], [[TMP41]]
-// CHECK:   store volatile <2 x i64> [[SUB20]], ptr @sl, align 8
-// CHECK:   [[TMP42:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP43:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[SUB21:%.*]] = sub <2 x i64> [[TMP42]], [[TMP43]]
-// CHECK:   store volatile <2 x i64> [[SUB21]], ptr @ul, align 8
-// CHECK:   [[TMP44:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP45:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[SUB22:%.*]] = sub <2 x i64> [[TMP44]], [[TMP45]]
-// CHECK:   store volatile <2 x i64> [[SUB22]], ptr @ul, align 8
-// CHECK:   [[TMP46:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP47:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[SUB23:%.*]] = sub <2 x i64> [[TMP46]], [[TMP47]]
-// CHECK:   store volatile <2 x i64> [[SUB23]], ptr @ul, align 8
-// CHECK:   [[TMP48:%.*]] = load volatile <2 x double>, ptr @fd, align 8
-// CHECK:   [[TMP49:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[SUB24:%.*]] = fsub <2 x double> [[TMP48]], [[TMP49]]
-// CHECK:   store volatile <2 x double> [[SUB24]], ptr @fd, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_sub(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[SUB:%.*]] = sub <16 x i8> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SUB]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[SUB1:%.*]] = sub <16 x i8> [[TMP2]], [[TMP3]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SUB1]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[SUB2:%.*]] = sub <16 x i8> [[TMP4]], [[TMP5]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SUB2]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[SUB3:%.*]] = sub <16 x i8> [[TMP6]], [[TMP7]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SUB3]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[SUB4:%.*]] = sub <16 x i8> [[TMP8]], [[TMP9]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SUB4]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[SUB5:%.*]] = sub <16 x i8> [[TMP10]], [[TMP11]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SUB5]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[SUB6:%.*]] = sub <8 x i16> [[TMP12]], [[TMP13]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SUB6]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[SUB7:%.*]] = sub <8 x i16> [[TMP14]], [[TMP15]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SUB7]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[SUB8:%.*]] = sub <8 x i16> [[TMP16]], [[TMP17]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SUB8]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[SUB9:%.*]] = sub <8 x i16> [[TMP18]], [[TMP19]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SUB9]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[SUB10:%.*]] = sub <8 x i16> [[TMP20]], [[TMP21]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SUB10]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[SUB11:%.*]] = sub <8 x i16> [[TMP22]], [[TMP23]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SUB11]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[SUB12:%.*]] = sub <4 x i32> [[TMP24]], [[TMP25]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SUB12]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP26:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP27:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[SUB13:%.*]] = sub <4 x i32> [[TMP26]], [[TMP27]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SUB13]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP28:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[SUB14:%.*]] = sub <4 x i32> [[TMP28]], [[TMP29]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SUB14]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP30:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP31:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[SUB15:%.*]] = sub <4 x i32> [[TMP30]], [[TMP31]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SUB15]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP32:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP33:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[SUB16:%.*]] = sub <4 x i32> [[TMP32]], [[TMP33]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SUB16]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP34:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP35:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[SUB17:%.*]] = sub <4 x i32> [[TMP34]], [[TMP35]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SUB17]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP36:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP37:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[SUB18:%.*]] = sub <2 x i64> [[TMP36]], [[TMP37]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SUB18]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP38:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP39:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[SUB19:%.*]] = sub <2 x i64> [[TMP38]], [[TMP39]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SUB19]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP40:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP41:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[SUB20:%.*]] = sub <2 x i64> [[TMP40]], [[TMP41]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SUB20]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP42:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP43:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[SUB21:%.*]] = sub <2 x i64> [[TMP42]], [[TMP43]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SUB21]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP44:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP45:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[SUB22:%.*]] = sub <2 x i64> [[TMP44]], [[TMP45]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SUB22]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP46:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP47:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[SUB23:%.*]] = sub <2 x i64> [[TMP46]], [[TMP47]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SUB23]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP48:%.*]] = load volatile <2 x double>, ptr @fd, align 8
+// CHECK-NEXT:    [[TMP49:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[SUB24:%.*]] = fsub <2 x double> [[TMP48]], [[TMP49]]
+// CHECK-NEXT:    store volatile <2 x double> [[SUB24]], ptr @fd, align 8
+// CHECK-NEXT:    ret void
+//
 void test_sub(void) {
 
   sc = sc - sc2;
@@ -671,76 +702,79 @@ void test_sub(void) {
   fd = fd - fd2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_sub_assign() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[SUB:%.*]] = sub <16 x i8> [[TMP1]], [[TMP0]]
-// CHECK:   store volatile <16 x i8> [[SUB]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[SUB1:%.*]] = sub <16 x i8> [[TMP3]], [[TMP2]]
-// CHECK:   store volatile <16 x i8> [[SUB1]], ptr @sc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[SUB2:%.*]] = sub <16 x i8> [[TMP5]], [[TMP4]]
-// CHECK:   store volatile <16 x i8> [[SUB2]], ptr @uc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[SUB3:%.*]] = sub <16 x i8> [[TMP7]], [[TMP6]]
-// CHECK:   store volatile <16 x i8> [[SUB3]], ptr @uc, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[SUB4:%.*]] = sub <8 x i16> [[TMP9]], [[TMP8]]
-// CHECK:   store volatile <8 x i16> [[SUB4]], ptr @ss, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[SUB5:%.*]] = sub <8 x i16> [[TMP11]], [[TMP10]]
-// CHECK:   store volatile <8 x i16> [[SUB5]], ptr @ss, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[SUB6:%.*]] = sub <8 x i16> [[TMP13]], [[TMP12]]
-// CHECK:   store volatile <8 x i16> [[SUB6]], ptr @us, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[SUB7:%.*]] = sub <8 x i16> [[TMP15]], [[TMP14]]
-// CHECK:   store volatile <8 x i16> [[SUB7]], ptr @us, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[SUB8:%.*]] = sub <4 x i32> [[TMP17]], [[TMP16]]
-// CHECK:   store volatile <4 x i32> [[SUB8]], ptr @si, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[TMP19:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[SUB9:%.*]] = sub <4 x i32> [[TMP19]], [[TMP18]]
-// CHECK:   store volatile <4 x i32> [[SUB9]], ptr @si, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[SUB10:%.*]] = sub <4 x i32> [[TMP21]], [[TMP20]]
-// CHECK:   store volatile <4 x i32> [[SUB10]], ptr @ui, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[SUB11:%.*]] = sub <4 x i32> [[TMP23]], [[TMP22]]
-// CHECK:   store volatile <4 x i32> [[SUB11]], ptr @ui, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[SUB12:%.*]] = sub <2 x i64> [[TMP25]], [[TMP24]]
-// CHECK:   store volatile <2 x i64> [[SUB12]], ptr @sl, align 8
-// CHECK:   [[TMP26:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[TMP27:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[SUB13:%.*]] = sub <2 x i64> [[TMP27]], [[TMP26]]
-// CHECK:   store volatile <2 x i64> [[SUB13]], ptr @sl, align 8
-// CHECK:   [[TMP28:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[TMP29:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[SUB14:%.*]] = sub <2 x i64> [[TMP29]], [[TMP28]]
-// CHECK:   store volatile <2 x i64> [[SUB14]], ptr @ul, align 8
-// CHECK:   [[TMP30:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[TMP31:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[SUB15:%.*]] = sub <2 x i64> [[TMP31]], [[TMP30]]
-// CHECK:   store volatile <2 x i64> [[SUB15]], ptr @ul, align 8
-// CHECK:   [[TMP32:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[TMP33:%.*]] = load volatile <2 x double>, ptr @fd, align 8
-// CHECK:   [[SUB16:%.*]] = fsub <2 x double> [[TMP33]], [[TMP32]]
-// CHECK:   store volatile <2 x double> [[SUB16]], ptr @fd, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_sub_assign(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[SUB:%.*]] = sub <16 x i8> [[TMP1]], [[TMP0]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SUB]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[SUB1:%.*]] = sub <16 x i8> [[TMP3]], [[TMP2]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SUB1]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[SUB2:%.*]] = sub <16 x i8> [[TMP5]], [[TMP4]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SUB2]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[SUB3:%.*]] = sub <16 x i8> [[TMP7]], [[TMP6]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SUB3]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[SUB4:%.*]] = sub <8 x i16> [[TMP9]], [[TMP8]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SUB4]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[SUB5:%.*]] = sub <8 x i16> [[TMP11]], [[TMP10]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SUB5]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[SUB6:%.*]] = sub <8 x i16> [[TMP13]], [[TMP12]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SUB6]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[SUB7:%.*]] = sub <8 x i16> [[TMP15]], [[TMP14]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SUB7]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[SUB8:%.*]] = sub <4 x i32> [[TMP17]], [[TMP16]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SUB8]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[SUB9:%.*]] = sub <4 x i32> [[TMP19]], [[TMP18]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SUB9]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[SUB10:%.*]] = sub <4 x i32> [[TMP21]], [[TMP20]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SUB10]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[SUB11:%.*]] = sub <4 x i32> [[TMP23]], [[TMP22]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SUB11]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[SUB12:%.*]] = sub <2 x i64> [[TMP25]], [[TMP24]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SUB12]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP26:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[TMP27:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[SUB13:%.*]] = sub <2 x i64> [[TMP27]], [[TMP26]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SUB13]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP28:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[TMP29:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[SUB14:%.*]] = sub <2 x i64> [[TMP29]], [[TMP28]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SUB14]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP30:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[TMP31:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[SUB15:%.*]] = sub <2 x i64> [[TMP31]], [[TMP30]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SUB15]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP32:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[TMP33:%.*]] = load volatile <2 x double>, ptr @fd, align 8
+// CHECK-NEXT:    [[SUB16:%.*]] = fsub <2 x double> [[TMP33]], [[TMP32]]
+// CHECK-NEXT:    store volatile <2 x double> [[SUB16]], ptr @fd, align 8
+// CHECK-NEXT:    ret void
+//
 void test_sub_assign(void) {
 
   sc -= sc2;
@@ -766,44 +800,47 @@ void test_sub_assign(void) {
   fd -= fd2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_mul() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[MUL:%.*]] = mul <16 x i8> [[TMP0]], [[TMP1]]
-// CHECK:   store volatile <16 x i8> [[MUL]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[MUL1:%.*]] = mul <16 x i8> [[TMP2]], [[TMP3]]
-// CHECK:   store volatile <16 x i8> [[MUL1]], ptr @uc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[MUL2:%.*]] = mul <8 x i16> [[TMP4]], [[TMP5]]
-// CHECK:   store volatile <8 x i16> [[MUL2]], ptr @ss, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[MUL3:%.*]] = mul <8 x i16> [[TMP6]], [[TMP7]]
-// CHECK:   store volatile <8 x i16> [[MUL3]], ptr @us, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[MUL4:%.*]] = mul <4 x i32> [[TMP8]], [[TMP9]]
-// CHECK:   store volatile <4 x i32> [[MUL4]], ptr @si, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[MUL5:%.*]] = mul <4 x i32> [[TMP10]], [[TMP11]]
-// CHECK:   store volatile <4 x i32> [[MUL5]], ptr @ui, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[MUL6:%.*]] = mul <2 x i64> [[TMP12]], [[TMP13]]
-// CHECK:   store volatile <2 x i64> [[MUL6]], ptr @sl, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[MUL7:%.*]] = mul <2 x i64> [[TMP14]], [[TMP15]]
-// CHECK:   store volatile <2 x i64> [[MUL7]], ptr @ul, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <2 x double>, ptr @fd, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[MUL8:%.*]] = fmul <2 x double> [[TMP16]], [[TMP17]]
-// CHECK:   store volatile <2 x double> [[MUL8]], ptr @fd, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_mul(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[MUL:%.*]] = mul <16 x i8> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    store volatile <16 x i8> [[MUL]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[MUL1:%.*]] = mul <16 x i8> [[TMP2]], [[TMP3]]
+// CHECK-NEXT:    store volatile <16 x i8> [[MUL1]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[MUL2:%.*]] = mul <8 x i16> [[TMP4]], [[TMP5]]
+// CHECK-NEXT:    store volatile <8 x i16> [[MUL2]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[MUL3:%.*]] = mul <8 x i16> [[TMP6]], [[TMP7]]
+// CHECK-NEXT:    store volatile <8 x i16> [[MUL3]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[MUL4:%.*]] = mul <4 x i32> [[TMP8]], [[TMP9]]
+// CHECK-NEXT:    store volatile <4 x i32> [[MUL4]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[MUL5:%.*]] = mul <4 x i32> [[TMP10]], [[TMP11]]
+// CHECK-NEXT:    store volatile <4 x i32> [[MUL5]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[MUL6:%.*]] = mul <2 x i64> [[TMP12]], [[TMP13]]
+// CHECK-NEXT:    store volatile <2 x i64> [[MUL6]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[MUL7:%.*]] = mul <2 x i64> [[TMP14]], [[TMP15]]
+// CHECK-NEXT:    store volatile <2 x i64> [[MUL7]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <2 x double>, ptr @fd, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[MUL8:%.*]] = fmul <2 x double> [[TMP16]], [[TMP17]]
+// CHECK-NEXT:    store volatile <2 x double> [[MUL8]], ptr @fd, align 8
+// CHECK-NEXT:    ret void
+//
 void test_mul(void) {
 
   sc = sc * sc2;
@@ -821,44 +858,47 @@ void test_mul(void) {
   fd = fd * fd2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_mul_assign() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[MUL:%.*]] = mul <16 x i8> [[TMP1]], [[TMP0]]
-// CHECK:   store volatile <16 x i8> [[MUL]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[MUL1:%.*]] = mul <16 x i8> [[TMP3]], [[TMP2]]
-// CHECK:   store volatile <16 x i8> [[MUL1]], ptr @uc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[MUL2:%.*]] = mul <8 x i16> [[TMP5]], [[TMP4]]
-// CHECK:   store volatile <8 x i16> [[MUL2]], ptr @ss, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[MUL3:%.*]] = mul <8 x i16> [[TMP7]], [[TMP6]]
-// CHECK:   store volatile <8 x i16> [[MUL3]], ptr @us, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[MUL4:%.*]] = mul <4 x i32> [[TMP9]], [[TMP8]]
-// CHECK:   store volatile <4 x i32> [[MUL4]], ptr @si, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[MUL5:%.*]] = mul <4 x i32> [[TMP11]], [[TMP10]]
-// CHECK:   store volatile <4 x i32> [[MUL5]], ptr @ui, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[MUL6:%.*]] = mul <2 x i64> [[TMP13]], [[TMP12]]
-// CHECK:   store volatile <2 x i64> [[MUL6]], ptr @sl, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[MUL7:%.*]] = mul <2 x i64> [[TMP15]], [[TMP14]]
-// CHECK:   store volatile <2 x i64> [[MUL7]], ptr @ul, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <2 x double>, ptr @fd, align 8
-// CHECK:   [[MUL8:%.*]] = fmul <2 x double> [[TMP17]], [[TMP16]]
-// CHECK:   store volatile <2 x double> [[MUL8]], ptr @fd, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_mul_assign(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[MUL:%.*]] = mul <16 x i8> [[TMP1]], [[TMP0]]
+// CHECK-NEXT:    store volatile <16 x i8> [[MUL]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[MUL1:%.*]] = mul <16 x i8> [[TMP3]], [[TMP2]]
+// CHECK-NEXT:    store volatile <16 x i8> [[MUL1]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[MUL2:%.*]] = mul <8 x i16> [[TMP5]], [[TMP4]]
+// CHECK-NEXT:    store volatile <8 x i16> [[MUL2]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[MUL3:%.*]] = mul <8 x i16> [[TMP7]], [[TMP6]]
+// CHECK-NEXT:    store volatile <8 x i16> [[MUL3]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[MUL4:%.*]] = mul <4 x i32> [[TMP9]], [[TMP8]]
+// CHECK-NEXT:    store volatile <4 x i32> [[MUL4]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[MUL5:%.*]] = mul <4 x i32> [[TMP11]], [[TMP10]]
+// CHECK-NEXT:    store volatile <4 x i32> [[MUL5]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[MUL6:%.*]] = mul <2 x i64> [[TMP13]], [[TMP12]]
+// CHECK-NEXT:    store volatile <2 x i64> [[MUL6]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[MUL7:%.*]] = mul <2 x i64> [[TMP15]], [[TMP14]]
+// CHECK-NEXT:    store volatile <2 x i64> [[MUL7]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <2 x double>, ptr @fd, align 8
+// CHECK-NEXT:    [[MUL8:%.*]] = fmul <2 x double> [[TMP17]], [[TMP16]]
+// CHECK-NEXT:    store volatile <2 x double> [[MUL8]], ptr @fd, align 8
+// CHECK-NEXT:    ret void
+//
 void test_mul_assign(void) {
 
   sc *= sc2;
@@ -876,44 +916,47 @@ void test_mul_assign(void) {
   fd *= fd2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_div() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[DIV:%.*]] = sdiv <16 x i8> [[TMP0]], [[TMP1]]
-// CHECK:   store volatile <16 x i8> [[DIV]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[DIV1:%.*]] = udiv <16 x i8> [[TMP2]], [[TMP3]]
-// CHECK:   store volatile <16 x i8> [[DIV1]], ptr @uc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[DIV2:%.*]] = sdiv <8 x i16> [[TMP4]], [[TMP5]]
-// CHECK:   store volatile <8 x i16> [[DIV2]], ptr @ss, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[DIV3:%.*]] = udiv <8 x i16> [[TMP6]], [[TMP7]]
-// CHECK:   store volatile <8 x i16> [[DIV3]], ptr @us, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[DIV4:%.*]] = sdiv <4 x i32> [[TMP8]], [[TMP9]]
-// CHECK:   store volatile <4 x i32> [[DIV4]], ptr @si, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[DIV5:%.*]] = udiv <4 x i32> [[TMP10]], [[TMP11]]
-// CHECK:   store volatile <4 x i32> [[DIV5]], ptr @ui, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[DIV6:%.*]] = sdiv <2 x i64> [[TMP12]], [[TMP13]]
-// CHECK:   store volatile <2 x i64> [[DIV6]], ptr @sl, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[DIV7:%.*]] = udiv <2 x i64> [[TMP14]], [[TMP15]]
-// CHECK:   store volatile <2 x i64> [[DIV7]], ptr @ul, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <2 x double>, ptr @fd, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[DIV8:%.*]] = fdiv <2 x double> [[TMP16]], [[TMP17]]
-// CHECK:   store volatile <2 x double> [[DIV8]], ptr @fd, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_div(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[DIV:%.*]] = sdiv <16 x i8> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    store volatile <16 x i8> [[DIV]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[DIV1:%.*]] = udiv <16 x i8> [[TMP2]], [[TMP3]]
+// CHECK-NEXT:    store volatile <16 x i8> [[DIV1]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[DIV2:%.*]] = sdiv <8 x i16> [[TMP4]], [[TMP5]]
+// CHECK-NEXT:    store volatile <8 x i16> [[DIV2]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[DIV3:%.*]] = udiv <8 x i16> [[TMP6]], [[TMP7]]
+// CHECK-NEXT:    store volatile <8 x i16> [[DIV3]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[DIV4:%.*]] = sdiv <4 x i32> [[TMP8]], [[TMP9]]
+// CHECK-NEXT:    store volatile <4 x i32> [[DIV4]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[DIV5:%.*]] = udiv <4 x i32> [[TMP10]], [[TMP11]]
+// CHECK-NEXT:    store volatile <4 x i32> [[DIV5]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[DIV6:%.*]] = sdiv <2 x i64> [[TMP12]], [[TMP13]]
+// CHECK-NEXT:    store volatile <2 x i64> [[DIV6]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[DIV7:%.*]] = udiv <2 x i64> [[TMP14]], [[TMP15]]
+// CHECK-NEXT:    store volatile <2 x i64> [[DIV7]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <2 x double>, ptr @fd, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[DIV8:%.*]] = fdiv <2 x double> [[TMP16]], [[TMP17]]
+// CHECK-NEXT:    store volatile <2 x double> [[DIV8]], ptr @fd, align 8
+// CHECK-NEXT:    ret void
+//
 void test_div(void) {
 
   sc = sc / sc2;
@@ -931,44 +974,47 @@ void test_div(void) {
   fd = fd / fd2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_div_assign() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[DIV:%.*]] = sdiv <16 x i8> [[TMP1]], [[TMP0]]
-// CHECK:   store volatile <16 x i8> [[DIV]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[DIV1:%.*]] = udiv <16 x i8> [[TMP3]], [[TMP2]]
-// CHECK:   store volatile <16 x i8> [[DIV1]], ptr @uc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[DIV2:%.*]] = sdiv <8 x i16> [[TMP5]], [[TMP4]]
-// CHECK:   store volatile <8 x i16> [[DIV2]], ptr @ss, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[DIV3:%.*]] = udiv <8 x i16> [[TMP7]], [[TMP6]]
-// CHECK:   store volatile <8 x i16> [[DIV3]], ptr @us, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[DIV4:%.*]] = sdiv <4 x i32> [[TMP9]], [[TMP8]]
-// CHECK:   store volatile <4 x i32> [[DIV4]], ptr @si, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[DIV5:%.*]] = udiv <4 x i32> [[TMP11]], [[TMP10]]
-// CHECK:   store volatile <4 x i32> [[DIV5]], ptr @ui, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[DIV6:%.*]] = sdiv <2 x i64> [[TMP13]], [[TMP12]]
-// CHECK:   store volatile <2 x i64> [[DIV6]], ptr @sl, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[DIV7:%.*]] = udiv <2 x i64> [[TMP15]], [[TMP14]]
-// CHECK:   store volatile <2 x i64> [[DIV7]], ptr @ul, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <2 x double>, ptr @fd, align 8
-// CHECK:   [[DIV8:%.*]] = fdiv <2 x double> [[TMP17]], [[TMP16]]
-// CHECK:   store volatile <2 x double> [[DIV8]], ptr @fd, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_div_assign(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[DIV:%.*]] = sdiv <16 x i8> [[TMP1]], [[TMP0]]
+// CHECK-NEXT:    store volatile <16 x i8> [[DIV]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[DIV1:%.*]] = udiv <16 x i8> [[TMP3]], [[TMP2]]
+// CHECK-NEXT:    store volatile <16 x i8> [[DIV1]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[DIV2:%.*]] = sdiv <8 x i16> [[TMP5]], [[TMP4]]
+// CHECK-NEXT:    store volatile <8 x i16> [[DIV2]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[DIV3:%.*]] = udiv <8 x i16> [[TMP7]], [[TMP6]]
+// CHECK-NEXT:    store volatile <8 x i16> [[DIV3]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[DIV4:%.*]] = sdiv <4 x i32> [[TMP9]], [[TMP8]]
+// CHECK-NEXT:    store volatile <4 x i32> [[DIV4]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[DIV5:%.*]] = udiv <4 x i32> [[TMP11]], [[TMP10]]
+// CHECK-NEXT:    store volatile <4 x i32> [[DIV5]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[DIV6:%.*]] = sdiv <2 x i64> [[TMP13]], [[TMP12]]
+// CHECK-NEXT:    store volatile <2 x i64> [[DIV6]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[DIV7:%.*]] = udiv <2 x i64> [[TMP15]], [[TMP14]]
+// CHECK-NEXT:    store volatile <2 x i64> [[DIV7]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <2 x double>, ptr @fd, align 8
+// CHECK-NEXT:    [[DIV8:%.*]] = fdiv <2 x double> [[TMP17]], [[TMP16]]
+// CHECK-NEXT:    store volatile <2 x double> [[DIV8]], ptr @fd, align 8
+// CHECK-NEXT:    ret void
+//
 void test_div_assign(void) {
 
   sc /= sc2;
@@ -986,40 +1032,43 @@ void test_div_assign(void) {
   fd /= fd2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_rem() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[REM:%.*]] = srem <16 x i8> [[TMP0]], [[TMP1]]
-// CHECK:   store volatile <16 x i8> [[REM]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[REM1:%.*]] = urem <16 x i8> [[TMP2]], [[TMP3]]
-// CHECK:   store volatile <16 x i8> [[REM1]], ptr @uc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[REM2:%.*]] = srem <8 x i16> [[TMP4]], [[TMP5]]
-// CHECK:   store volatile <8 x i16> [[REM2]], ptr @ss, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[REM3:%.*]] = urem <8 x i16> [[TMP6]], [[TMP7]]
-// CHECK:   store volatile <8 x i16> [[REM3]], ptr @us, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[REM4:%.*]] = srem <4 x i32> [[TMP8]], [[TMP9]]
-// CHECK:   store volatile <4 x i32> [[REM4]], ptr @si, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[REM5:%.*]] = urem <4 x i32> [[TMP10]], [[TMP11]]
-// CHECK:   store volatile <4 x i32> [[REM5]], ptr @ui, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[REM6:%.*]] = srem <2 x i64> [[TMP12]], [[TMP13]]
-// CHECK:   store volatile <2 x i64> [[REM6]], ptr @sl, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[REM7:%.*]] = urem <2 x i64> [[TMP14]], [[TMP15]]
-// CHECK:   store volatile <2 x i64> [[REM7]], ptr @ul, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_rem(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[REM:%.*]] = srem <16 x i8> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    store volatile <16 x i8> [[REM]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[REM1:%.*]] = urem <16 x i8> [[TMP2]], [[TMP3]]
+// CHECK-NEXT:    store volatile <16 x i8> [[REM1]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[REM2:%.*]] = srem <8 x i16> [[TMP4]], [[TMP5]]
+// CHECK-NEXT:    store volatile <8 x i16> [[REM2]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[REM3:%.*]] = urem <8 x i16> [[TMP6]], [[TMP7]]
+// CHECK-NEXT:    store volatile <8 x i16> [[REM3]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[REM4:%.*]] = srem <4 x i32> [[TMP8]], [[TMP9]]
+// CHECK-NEXT:    store volatile <4 x i32> [[REM4]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[REM5:%.*]] = urem <4 x i32> [[TMP10]], [[TMP11]]
+// CHECK-NEXT:    store volatile <4 x i32> [[REM5]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[REM6:%.*]] = srem <2 x i64> [[TMP12]], [[TMP13]]
+// CHECK-NEXT:    store volatile <2 x i64> [[REM6]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[REM7:%.*]] = urem <2 x i64> [[TMP14]], [[TMP15]]
+// CHECK-NEXT:    store volatile <2 x i64> [[REM7]], ptr @ul, align 8
+// CHECK-NEXT:    ret void
+//
 void test_rem(void) {
 
   sc = sc % sc2;
@@ -1035,40 +1084,43 @@ void test_rem(void) {
   ul = ul % ul2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_rem_assign() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[REM:%.*]] = srem <16 x i8> [[TMP1]], [[TMP0]]
-// CHECK:   store volatile <16 x i8> [[REM]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[REM1:%.*]] = urem <16 x i8> [[TMP3]], [[TMP2]]
-// CHECK:   store volatile <16 x i8> [[REM1]], ptr @uc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[REM2:%.*]] = srem <8 x i16> [[TMP5]], [[TMP4]]
-// CHECK:   store volatile <8 x i16> [[REM2]], ptr @ss, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[REM3:%.*]] = urem <8 x i16> [[TMP7]], [[TMP6]]
-// CHECK:   store volatile <8 x i16> [[REM3]], ptr @us, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[REM4:%.*]] = srem <4 x i32> [[TMP9]], [[TMP8]]
-// CHECK:   store volatile <4 x i32> [[REM4]], ptr @si, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[REM5:%.*]] = urem <4 x i32> [[TMP11]], [[TMP10]]
-// CHECK:   store volatile <4 x i32> [[REM5]], ptr @ui, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[REM6:%.*]] = srem <2 x i64> [[TMP13]], [[TMP12]]
-// CHECK:   store volatile <2 x i64> [[REM6]], ptr @sl, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[REM7:%.*]] = urem <2 x i64> [[TMP15]], [[TMP14]]
-// CHECK:   store volatile <2 x i64> [[REM7]], ptr @ul, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_rem_assign(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[REM:%.*]] = srem <16 x i8> [[TMP1]], [[TMP0]]
+// CHECK-NEXT:    store volatile <16 x i8> [[REM]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[REM1:%.*]] = urem <16 x i8> [[TMP3]], [[TMP2]]
+// CHECK-NEXT:    store volatile <16 x i8> [[REM1]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[REM2:%.*]] = srem <8 x i16> [[TMP5]], [[TMP4]]
+// CHECK-NEXT:    store volatile <8 x i16> [[REM2]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[REM3:%.*]] = urem <8 x i16> [[TMP7]], [[TMP6]]
+// CHECK-NEXT:    store volatile <8 x i16> [[REM3]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[REM4:%.*]] = srem <4 x i32> [[TMP9]], [[TMP8]]
+// CHECK-NEXT:    store volatile <4 x i32> [[REM4]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[REM5:%.*]] = urem <4 x i32> [[TMP11]], [[TMP10]]
+// CHECK-NEXT:    store volatile <4 x i32> [[REM5]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[REM6:%.*]] = srem <2 x i64> [[TMP13]], [[TMP12]]
+// CHECK-NEXT:    store volatile <2 x i64> [[REM6]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[REM7:%.*]] = urem <2 x i64> [[TMP15]], [[TMP14]]
+// CHECK-NEXT:    store volatile <2 x i64> [[REM7]], ptr @ul, align 8
+// CHECK-NEXT:    ret void
+//
 void test_rem_assign(void) {
 
   sc %= sc2;
@@ -1084,44 +1136,47 @@ void test_rem_assign(void) {
   ul %= ul2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_not() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[NEG:%.*]] = xor <16 x i8> [[TMP0]], splat (i8 -1)
-// CHECK:   store volatile <16 x i8> [[NEG]], ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[NEG1:%.*]] = xor <16 x i8> [[TMP1]], splat (i8 -1)
-// CHECK:   store volatile <16 x i8> [[NEG1]], ptr @uc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[NEG2:%.*]] = xor <16 x i8> [[TMP2]], splat (i8 -1)
-// CHECK:   store volatile <16 x i8> [[NEG2]], ptr @bc, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[NEG3:%.*]] = xor <8 x i16> [[TMP3]], splat (i16 -1)
-// CHECK:   store volatile <8 x i16> [[NEG3]], ptr @ss, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[NEG4:%.*]] = xor <8 x i16> [[TMP4]], splat (i16 -1)
-// CHECK:   store volatile <8 x i16> [[NEG4]], ptr @us, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[NEG5:%.*]] = xor <8 x i16> [[TMP5]], splat (i16 -1)
-// CHECK:   store volatile <8 x i16> [[NEG5]], ptr @bs, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[NEG6:%.*]] = xor <4 x i32> [[TMP6]], splat (i32 -1)
-// CHECK:   store volatile <4 x i32> [[NEG6]], ptr @si, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[NEG7:%.*]] = xor <4 x i32> [[TMP7]], splat (i32 -1)
-// CHECK:   store volatile <4 x i32> [[NEG7]], ptr @ui, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[NEG8:%.*]] = xor <4 x i32> [[TMP8]], splat (i32 -1)
-// CHECK:   store volatile <4 x i32> [[NEG8]], ptr @bi, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[NEG9:%.*]] = xor <2 x i64> [[TMP9]], splat (i64 -1)
-// CHECK:   store volatile <2 x i64> [[NEG9]], ptr @sl, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[NEG10:%.*]] = xor <2 x i64> [[TMP10]], splat (i64 -1)
-// CHECK:   store volatile <2 x i64> [[NEG10]], ptr @ul, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[NEG11:%.*]] = xor <2 x i64> [[TMP11]], splat (i64 -1)
-// CHECK:   store volatile <2 x i64> [[NEG11]], ptr @bl, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_not(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[NOT:%.*]] = xor <16 x i8> [[TMP0]], splat (i8 -1)
+// CHECK-NEXT:    store volatile <16 x i8> [[NOT]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[NOT1:%.*]] = xor <16 x i8> [[TMP1]], splat (i8 -1)
+// CHECK-NEXT:    store volatile <16 x i8> [[NOT1]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[NOT2:%.*]] = xor <16 x i8> [[TMP2]], splat (i8 -1)
+// CHECK-NEXT:    store volatile <16 x i8> [[NOT2]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[NOT3:%.*]] = xor <8 x i16> [[TMP3]], splat (i16 -1)
+// CHECK-NEXT:    store volatile <8 x i16> [[NOT3]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[NOT4:%.*]] = xor <8 x i16> [[TMP4]], splat (i16 -1)
+// CHECK-NEXT:    store volatile <8 x i16> [[NOT4]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[NOT5:%.*]] = xor <8 x i16> [[TMP5]], splat (i16 -1)
+// CHECK-NEXT:    store volatile <8 x i16> [[NOT5]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[NOT6:%.*]] = xor <4 x i32> [[TMP6]], splat (i32 -1)
+// CHECK-NEXT:    store volatile <4 x i32> [[NOT6]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[NOT7:%.*]] = xor <4 x i32> [[TMP7]], splat (i32 -1)
+// CHECK-NEXT:    store volatile <4 x i32> [[NOT7]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[NOT8:%.*]] = xor <4 x i32> [[TMP8]], splat (i32 -1)
+// CHECK-NEXT:    store volatile <4 x i32> [[NOT8]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[NOT9:%.*]] = xor <2 x i64> [[TMP9]], splat (i64 -1)
+// CHECK-NEXT:    store volatile <2 x i64> [[NOT9]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[NOT10:%.*]] = xor <2 x i64> [[TMP10]], splat (i64 -1)
+// CHECK-NEXT:    store volatile <2 x i64> [[NOT10]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[NOT11:%.*]] = xor <2 x i64> [[TMP11]], splat (i64 -1)
+// CHECK-NEXT:    store volatile <2 x i64> [[NOT11]], ptr @bl, align 8
+// CHECK-NEXT:    ret void
+//
 void test_not(void) {
 
   sc = ~sc2;
@@ -1141,120 +1196,123 @@ void test_not(void) {
   bl = ~bl2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_and() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[AND:%.*]] = and <16 x i8> [[TMP0]], [[TMP1]]
-// CHECK:   store volatile <16 x i8> [[AND]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[AND1:%.*]] = and <16 x i8> [[TMP2]], [[TMP3]]
-// CHECK:   store volatile <16 x i8> [[AND1]], ptr @sc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[AND2:%.*]] = and <16 x i8> [[TMP4]], [[TMP5]]
-// CHECK:   store volatile <16 x i8> [[AND2]], ptr @sc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[AND3:%.*]] = and <16 x i8> [[TMP6]], [[TMP7]]
-// CHECK:   store volatile <16 x i8> [[AND3]], ptr @uc, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[AND4:%.*]] = and <16 x i8> [[TMP8]], [[TMP9]]
-// CHECK:   store volatile <16 x i8> [[AND4]], ptr @uc, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[AND5:%.*]] = and <16 x i8> [[TMP10]], [[TMP11]]
-// CHECK:   store volatile <16 x i8> [[AND5]], ptr @uc, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[AND6:%.*]] = and <16 x i8> [[TMP12]], [[TMP13]]
-// CHECK:   store volatile <16 x i8> [[AND6]], ptr @bc, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[AND7:%.*]] = and <8 x i16> [[TMP14]], [[TMP15]]
-// CHECK:   store volatile <8 x i16> [[AND7]], ptr @ss, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[AND8:%.*]] = and <8 x i16> [[TMP16]], [[TMP17]]
-// CHECK:   store volatile <8 x i16> [[AND8]], ptr @ss, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP19:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[AND9:%.*]] = and <8 x i16> [[TMP18]], [[TMP19]]
-// CHECK:   store volatile <8 x i16> [[AND9]], ptr @ss, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[AND10:%.*]] = and <8 x i16> [[TMP20]], [[TMP21]]
-// CHECK:   store volatile <8 x i16> [[AND10]], ptr @us, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[AND11:%.*]] = and <8 x i16> [[TMP22]], [[TMP23]]
-// CHECK:   store volatile <8 x i16> [[AND11]], ptr @us, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[AND12:%.*]] = and <8 x i16> [[TMP24]], [[TMP25]]
-// CHECK:   store volatile <8 x i16> [[AND12]], ptr @us, align 8
-// CHECK:   [[TMP26:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP27:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[AND13:%.*]] = and <8 x i16> [[TMP26]], [[TMP27]]
-// CHECK:   store volatile <8 x i16> [[AND13]], ptr @bs, align 8
-// CHECK:   [[TMP28:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[AND14:%.*]] = and <4 x i32> [[TMP28]], [[TMP29]]
-// CHECK:   store volatile <4 x i32> [[AND14]], ptr @si, align 8
-// CHECK:   [[TMP30:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP31:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[AND15:%.*]] = and <4 x i32> [[TMP30]], [[TMP31]]
-// CHECK:   store volatile <4 x i32> [[AND15]], ptr @si, align 8
-// CHECK:   [[TMP32:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP33:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[AND16:%.*]] = and <4 x i32> [[TMP32]], [[TMP33]]
-// CHECK:   store volatile <4 x i32> [[AND16]], ptr @si, align 8
-// CHECK:   [[TMP34:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP35:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[AND17:%.*]] = and <4 x i32> [[TMP34]], [[TMP35]]
-// CHECK:   store volatile <4 x i32> [[AND17]], ptr @ui, align 8
-// CHECK:   [[TMP36:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP37:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[AND18:%.*]] = and <4 x i32> [[TMP36]], [[TMP37]]
-// CHECK:   store volatile <4 x i32> [[AND18]], ptr @ui, align 8
-// CHECK:   [[TMP38:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP39:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[AND19:%.*]] = and <4 x i32> [[TMP38]], [[TMP39]]
-// CHECK:   store volatile <4 x i32> [[AND19]], ptr @ui, align 8
-// CHECK:   [[TMP40:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP41:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[AND20:%.*]] = and <4 x i32> [[TMP40]], [[TMP41]]
-// CHECK:   store volatile <4 x i32> [[AND20]], ptr @bi, align 8
-// CHECK:   [[TMP42:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP43:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[AND21:%.*]] = and <2 x i64> [[TMP42]], [[TMP43]]
-// CHECK:   store volatile <2 x i64> [[AND21]], ptr @sl, align 8
-// CHECK:   [[TMP44:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP45:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[AND22:%.*]] = and <2 x i64> [[TMP44]], [[TMP45]]
-// CHECK:   store volatile <2 x i64> [[AND22]], ptr @sl, align 8
-// CHECK:   [[TMP46:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP47:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[AND23:%.*]] = and <2 x i64> [[TMP46]], [[TMP47]]
-// CHECK:   store volatile <2 x i64> [[AND23]], ptr @sl, align 8
-// CHECK:   [[TMP48:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP49:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[AND24:%.*]] = and <2 x i64> [[TMP48]], [[TMP49]]
-// CHECK:   store volatile <2 x i64> [[AND24]], ptr @ul, align 8
-// CHECK:   [[TMP50:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP51:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[AND25:%.*]] = and <2 x i64> [[TMP50]], [[TMP51]]
-// CHECK:   store volatile <2 x i64> [[AND25]], ptr @ul, align 8
-// CHECK:   [[TMP52:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP53:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[AND26:%.*]] = and <2 x i64> [[TMP52]], [[TMP53]]
-// CHECK:   store volatile <2 x i64> [[AND26]], ptr @ul, align 8
-// CHECK:   [[TMP54:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP55:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[AND27:%.*]] = and <2 x i64> [[TMP54]], [[TMP55]]
-// CHECK:   store volatile <2 x i64> [[AND27]], ptr @bl, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_and(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[AND:%.*]] = and <16 x i8> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    store volatile <16 x i8> [[AND]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[AND1:%.*]] = and <16 x i8> [[TMP2]], [[TMP3]]
+// CHECK-NEXT:    store volatile <16 x i8> [[AND1]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[AND2:%.*]] = and <16 x i8> [[TMP4]], [[TMP5]]
+// CHECK-NEXT:    store volatile <16 x i8> [[AND2]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[AND3:%.*]] = and <16 x i8> [[TMP6]], [[TMP7]]
+// CHECK-NEXT:    store volatile <16 x i8> [[AND3]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[AND4:%.*]] = and <16 x i8> [[TMP8]], [[TMP9]]
+// CHECK-NEXT:    store volatile <16 x i8> [[AND4]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[AND5:%.*]] = and <16 x i8> [[TMP10]], [[TMP11]]
+// CHECK-NEXT:    store volatile <16 x i8> [[AND5]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[AND6:%.*]] = and <16 x i8> [[TMP12]], [[TMP13]]
+// CHECK-NEXT:    store volatile <16 x i8> [[AND6]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[AND7:%.*]] = and <8 x i16> [[TMP14]], [[TMP15]]
+// CHECK-NEXT:    store volatile <8 x i16> [[AND7]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[AND8:%.*]] = and <8 x i16> [[TMP16]], [[TMP17]]
+// CHECK-NEXT:    store volatile <8 x i16> [[AND8]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[AND9:%.*]] = and <8 x i16> [[TMP18]], [[TMP19]]
+// CHECK-NEXT:    store volatile <8 x i16> [[AND9]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[AND10:%.*]] = and <8 x i16> [[TMP20]], [[TMP21]]
+// CHECK-NEXT:    store volatile <8 x i16> [[AND10]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[AND11:%.*]] = and <8 x i16> [[TMP22]], [[TMP23]]
+// CHECK-NEXT:    store volatile <8 x i16> [[AND11]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[AND12:%.*]] = and <8 x i16> [[TMP24]], [[TMP25]]
+// CHECK-NEXT:    store volatile <8 x i16> [[AND12]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP26:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP27:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[AND13:%.*]] = and <8 x i16> [[TMP26]], [[TMP27]]
+// CHECK-NEXT:    store volatile <8 x i16> [[AND13]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP28:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[AND14:%.*]] = and <4 x i32> [[TMP28]], [[TMP29]]
+// CHECK-NEXT:    store volatile <4 x i32> [[AND14]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP30:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP31:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[AND15:%.*]] = and <4 x i32> [[TMP30]], [[TMP31]]
+// CHECK-NEXT:    store volatile <4 x i32> [[AND15]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP32:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP33:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[AND16:%.*]] = and <4 x i32> [[TMP32]], [[TMP33]]
+// CHECK-NEXT:    store volatile <4 x i32> [[AND16]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP34:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP35:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[AND17:%.*]] = and <4 x i32> [[TMP34]], [[TMP35]]
+// CHECK-NEXT:    store volatile <4 x i32> [[AND17]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP36:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP37:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[AND18:%.*]] = and <4 x i32> [[TMP36]], [[TMP37]]
+// CHECK-NEXT:    store volatile <4 x i32> [[AND18]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP38:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP39:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[AND19:%.*]] = and <4 x i32> [[TMP38]], [[TMP39]]
+// CHECK-NEXT:    store volatile <4 x i32> [[AND19]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP40:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP41:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[AND20:%.*]] = and <4 x i32> [[TMP40]], [[TMP41]]
+// CHECK-NEXT:    store volatile <4 x i32> [[AND20]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP42:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP43:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[AND21:%.*]] = and <2 x i64> [[TMP42]], [[TMP43]]
+// CHECK-NEXT:    store volatile <2 x i64> [[AND21]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP44:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP45:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[AND22:%.*]] = and <2 x i64> [[TMP44]], [[TMP45]]
+// CHECK-NEXT:    store volatile <2 x i64> [[AND22]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP46:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP47:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[AND23:%.*]] = and <2 x i64> [[TMP46]], [[TMP47]]
+// CHECK-NEXT:    store volatile <2 x i64> [[AND23]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP48:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP49:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[AND24:%.*]] = and <2 x i64> [[TMP48]], [[TMP49]]
+// CHECK-NEXT:    store volatile <2 x i64> [[AND24]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP50:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP51:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[AND25:%.*]] = and <2 x i64> [[TMP50]], [[TMP51]]
+// CHECK-NEXT:    store volatile <2 x i64> [[AND25]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP52:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP53:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[AND26:%.*]] = and <2 x i64> [[TMP52]], [[TMP53]]
+// CHECK-NEXT:    store volatile <2 x i64> [[AND26]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP54:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP55:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[AND27:%.*]] = and <2 x i64> [[TMP54]], [[TMP55]]
+// CHECK-NEXT:    store volatile <2 x i64> [[AND27]], ptr @bl, align 8
+// CHECK-NEXT:    ret void
+//
 void test_and(void) {
 
   sc = sc & sc2;
@@ -1290,88 +1348,91 @@ void test_and(void) {
   bl = bl & bl2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_and_assign() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[AND:%.*]] = and <16 x i8> [[TMP1]], [[TMP0]]
-// CHECK:   store volatile <16 x i8> [[AND]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[AND1:%.*]] = and <16 x i8> [[TMP3]], [[TMP2]]
-// CHECK:   store volatile <16 x i8> [[AND1]], ptr @sc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[AND2:%.*]] = and <16 x i8> [[TMP5]], [[TMP4]]
-// CHECK:   store volatile <16 x i8> [[AND2]], ptr @uc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[AND3:%.*]] = and <16 x i8> [[TMP7]], [[TMP6]]
-// CHECK:   store volatile <16 x i8> [[AND3]], ptr @uc, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[AND4:%.*]] = and <16 x i8> [[TMP9]], [[TMP8]]
-// CHECK:   store volatile <16 x i8> [[AND4]], ptr @bc, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[AND5:%.*]] = and <8 x i16> [[TMP11]], [[TMP10]]
-// CHECK:   store volatile <8 x i16> [[AND5]], ptr @ss, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[AND6:%.*]] = and <8 x i16> [[TMP13]], [[TMP12]]
-// CHECK:   store volatile <8 x i16> [[AND6]], ptr @ss, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[AND7:%.*]] = and <8 x i16> [[TMP15]], [[TMP14]]
-// CHECK:   store volatile <8 x i16> [[AND7]], ptr @us, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[AND8:%.*]] = and <8 x i16> [[TMP17]], [[TMP16]]
-// CHECK:   store volatile <8 x i16> [[AND8]], ptr @us, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[TMP19:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[AND9:%.*]] = and <8 x i16> [[TMP19]], [[TMP18]]
-// CHECK:   store volatile <8 x i16> [[AND9]], ptr @bs, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[AND10:%.*]] = and <4 x i32> [[TMP21]], [[TMP20]]
-// CHECK:   store volatile <4 x i32> [[AND10]], ptr @si, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[AND11:%.*]] = and <4 x i32> [[TMP23]], [[TMP22]]
-// CHECK:   store volatile <4 x i32> [[AND11]], ptr @si, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[AND12:%.*]] = and <4 x i32> [[TMP25]], [[TMP24]]
-// CHECK:   store volatile <4 x i32> [[AND12]], ptr @ui, align 8
-// CHECK:   [[TMP26:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[TMP27:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[AND13:%.*]] = and <4 x i32> [[TMP27]], [[TMP26]]
-// CHECK:   store volatile <4 x i32> [[AND13]], ptr @ui, align 8
-// CHECK:   [[TMP28:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[TMP29:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[AND14:%.*]] = and <4 x i32> [[TMP29]], [[TMP28]]
-// CHECK:   store volatile <4 x i32> [[AND14]], ptr @bi, align 8
-// CHECK:   [[TMP30:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[TMP31:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[AND15:%.*]] = and <2 x i64> [[TMP31]], [[TMP30]]
-// CHECK:   store volatile <2 x i64> [[AND15]], ptr @sl, align 8
-// CHECK:   [[TMP32:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[TMP33:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[AND16:%.*]] = and <2 x i64> [[TMP33]], [[TMP32]]
-// CHECK:   store volatile <2 x i64> [[AND16]], ptr @sl, align 8
-// CHECK:   [[TMP34:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[TMP35:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[AND17:%.*]] = and <2 x i64> [[TMP35]], [[TMP34]]
-// CHECK:   store volatile <2 x i64> [[AND17]], ptr @ul, align 8
-// CHECK:   [[TMP36:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[TMP37:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[AND18:%.*]] = and <2 x i64> [[TMP37]], [[TMP36]]
-// CHECK:   store volatile <2 x i64> [[AND18]], ptr @ul, align 8
-// CHECK:   [[TMP38:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[TMP39:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[AND19:%.*]] = and <2 x i64> [[TMP39]], [[TMP38]]
-// CHECK:   store volatile <2 x i64> [[AND19]], ptr @bl, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_and_assign(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[AND:%.*]] = and <16 x i8> [[TMP1]], [[TMP0]]
+// CHECK-NEXT:    store volatile <16 x i8> [[AND]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[AND1:%.*]] = and <16 x i8> [[TMP3]], [[TMP2]]
+// CHECK-NEXT:    store volatile <16 x i8> [[AND1]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[AND2:%.*]] = and <16 x i8> [[TMP5]], [[TMP4]]
+// CHECK-NEXT:    store volatile <16 x i8> [[AND2]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[AND3:%.*]] = and <16 x i8> [[TMP7]], [[TMP6]]
+// CHECK-NEXT:    store volatile <16 x i8> [[AND3]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[AND4:%.*]] = and <16 x i8> [[TMP9]], [[TMP8]]
+// CHECK-NEXT:    store volatile <16 x i8> [[AND4]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[AND5:%.*]] = and <8 x i16> [[TMP11]], [[TMP10]]
+// CHECK-NEXT:    store volatile <8 x i16> [[AND5]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[AND6:%.*]] = and <8 x i16> [[TMP13]], [[TMP12]]
+// CHECK-NEXT:    store volatile <8 x i16> [[AND6]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[AND7:%.*]] = and <8 x i16> [[TMP15]], [[TMP14]]
+// CHECK-NEXT:    store volatile <8 x i16> [[AND7]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[AND8:%.*]] = and <8 x i16> [[TMP17]], [[TMP16]]
+// CHECK-NEXT:    store volatile <8 x i16> [[AND8]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[AND9:%.*]] = and <8 x i16> [[TMP19]], [[TMP18]]
+// CHECK-NEXT:    store volatile <8 x i16> [[AND9]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[AND10:%.*]] = and <4 x i32> [[TMP21]], [[TMP20]]
+// CHECK-NEXT:    store volatile <4 x i32> [[AND10]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[AND11:%.*]] = and <4 x i32> [[TMP23]], [[TMP22]]
+// CHECK-NEXT:    store volatile <4 x i32> [[AND11]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[AND12:%.*]] = and <4 x i32> [[TMP25]], [[TMP24]]
+// CHECK-NEXT:    store volatile <4 x i32> [[AND12]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP26:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[TMP27:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[AND13:%.*]] = and <4 x i32> [[TMP27]], [[TMP26]]
+// CHECK-NEXT:    store volatile <4 x i32> [[AND13]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP28:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[TMP29:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[AND14:%.*]] = and <4 x i32> [[TMP29]], [[TMP28]]
+// CHECK-NEXT:    store volatile <4 x i32> [[AND14]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP30:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[TMP31:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[AND15:%.*]] = and <2 x i64> [[TMP31]], [[TMP30]]
+// CHECK-NEXT:    store volatile <2 x i64> [[AND15]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP32:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[TMP33:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[AND16:%.*]] = and <2 x i64> [[TMP33]], [[TMP32]]
+// CHECK-NEXT:    store volatile <2 x i64> [[AND16]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP34:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[TMP35:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[AND17:%.*]] = and <2 x i64> [[TMP35]], [[TMP34]]
+// CHECK-NEXT:    store volatile <2 x i64> [[AND17]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP36:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[TMP37:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[AND18:%.*]] = and <2 x i64> [[TMP37]], [[TMP36]]
+// CHECK-NEXT:    store volatile <2 x i64> [[AND18]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP38:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[TMP39:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[AND19:%.*]] = and <2 x i64> [[TMP39]], [[TMP38]]
+// CHECK-NEXT:    store volatile <2 x i64> [[AND19]], ptr @bl, align 8
+// CHECK-NEXT:    ret void
+//
 void test_and_assign(void) {
 
   sc &= sc2;
@@ -1399,120 +1460,123 @@ void test_and_assign(void) {
   bl &= bl2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_or() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[OR:%.*]] = or <16 x i8> [[TMP0]], [[TMP1]]
-// CHECK:   store volatile <16 x i8> [[OR]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[OR1:%.*]] = or <16 x i8> [[TMP2]], [[TMP3]]
-// CHECK:   store volatile <16 x i8> [[OR1]], ptr @sc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[OR2:%.*]] = or <16 x i8> [[TMP4]], [[TMP5]]
-// CHECK:   store volatile <16 x i8> [[OR2]], ptr @sc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[OR3:%.*]] = or <16 x i8> [[TMP6]], [[TMP7]]
-// CHECK:   store volatile <16 x i8> [[OR3]], ptr @uc, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[OR4:%.*]] = or <16 x i8> [[TMP8]], [[TMP9]]
-// CHECK:   store volatile <16 x i8> [[OR4]], ptr @uc, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[OR5:%.*]] = or <16 x i8> [[TMP10]], [[TMP11]]
-// CHECK:   store volatile <16 x i8> [[OR5]], ptr @uc, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[OR6:%.*]] = or <16 x i8> [[TMP12]], [[TMP13]]
-// CHECK:   store volatile <16 x i8> [[OR6]], ptr @bc, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[OR7:%.*]] = or <8 x i16> [[TMP14]], [[TMP15]]
-// CHECK:   store volatile <8 x i16> [[OR7]], ptr @ss, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[OR8:%.*]] = or <8 x i16> [[TMP16]], [[TMP17]]
-// CHECK:   store volatile <8 x i16> [[OR8]], ptr @ss, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP19:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[OR9:%.*]] = or <8 x i16> [[TMP18]], [[TMP19]]
-// CHECK:   store volatile <8 x i16> [[OR9]], ptr @ss, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[OR10:%.*]] = or <8 x i16> [[TMP20]], [[TMP21]]
-// CHECK:   store volatile <8 x i16> [[OR10]], ptr @us, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[OR11:%.*]] = or <8 x i16> [[TMP22]], [[TMP23]]
-// CHECK:   store volatile <8 x i16> [[OR11]], ptr @us, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[OR12:%.*]] = or <8 x i16> [[TMP24]], [[TMP25]]
-// CHECK:   store volatile <8 x i16> [[OR12]], ptr @us, align 8
-// CHECK:   [[TMP26:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP27:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[OR13:%.*]] = or <8 x i16> [[TMP26]], [[TMP27]]
-// CHECK:   store volatile <8 x i16> [[OR13]], ptr @bs, align 8
-// CHECK:   [[TMP28:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[OR14:%.*]] = or <4 x i32> [[TMP28]], [[TMP29]]
-// CHECK:   store volatile <4 x i32> [[OR14]], ptr @si, align 8
-// CHECK:   [[TMP30:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP31:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[OR15:%.*]] = or <4 x i32> [[TMP30]], [[TMP31]]
-// CHECK:   store volatile <4 x i32> [[OR15]], ptr @si, align 8
-// CHECK:   [[TMP32:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP33:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[OR16:%.*]] = or <4 x i32> [[TMP32]], [[TMP33]]
-// CHECK:   store volatile <4 x i32> [[OR16]], ptr @si, align 8
-// CHECK:   [[TMP34:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP35:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[OR17:%.*]] = or <4 x i32> [[TMP34]], [[TMP35]]
-// CHECK:   store volatile <4 x i32> [[OR17]], ptr @ui, align 8
-// CHECK:   [[TMP36:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP37:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[OR18:%.*]] = or <4 x i32> [[TMP36]], [[TMP37]]
-// CHECK:   store volatile <4 x i32> [[OR18]], ptr @ui, align 8
-// CHECK:   [[TMP38:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP39:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[OR19:%.*]] = or <4 x i32> [[TMP38]], [[TMP39]]
-// CHECK:   store volatile <4 x i32> [[OR19]], ptr @ui, align 8
-// CHECK:   [[TMP40:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP41:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[OR20:%.*]] = or <4 x i32> [[TMP40]], [[TMP41]]
-// CHECK:   store volatile <4 x i32> [[OR20]], ptr @bi, align 8
-// CHECK:   [[TMP42:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP43:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[OR21:%.*]] = or <2 x i64> [[TMP42]], [[TMP43]]
-// CHECK:   store volatile <2 x i64> [[OR21]], ptr @sl, align 8
-// CHECK:   [[TMP44:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP45:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[OR22:%.*]] = or <2 x i64> [[TMP44]], [[TMP45]]
-// CHECK:   store volatile <2 x i64> [[OR22]], ptr @sl, align 8
-// CHECK:   [[TMP46:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP47:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[OR23:%.*]] = or <2 x i64> [[TMP46]], [[TMP47]]
-// CHECK:   store volatile <2 x i64> [[OR23]], ptr @sl, align 8
-// CHECK:   [[TMP48:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP49:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[OR24:%.*]] = or <2 x i64> [[TMP48]], [[TMP49]]
-// CHECK:   store volatile <2 x i64> [[OR24]], ptr @ul, align 8
-// CHECK:   [[TMP50:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP51:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[OR25:%.*]] = or <2 x i64> [[TMP50]], [[TMP51]]
-// CHECK:   store volatile <2 x i64> [[OR25]], ptr @ul, align 8
-// CHECK:   [[TMP52:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP53:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[OR26:%.*]] = or <2 x i64> [[TMP52]], [[TMP53]]
-// CHECK:   store volatile <2 x i64> [[OR26]], ptr @ul, align 8
-// CHECK:   [[TMP54:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP55:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[OR27:%.*]] = or <2 x i64> [[TMP54]], [[TMP55]]
-// CHECK:   store volatile <2 x i64> [[OR27]], ptr @bl, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_or(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[OR:%.*]] = or <16 x i8> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    store volatile <16 x i8> [[OR]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[OR1:%.*]] = or <16 x i8> [[TMP2]], [[TMP3]]
+// CHECK-NEXT:    store volatile <16 x i8> [[OR1]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[OR2:%.*]] = or <16 x i8> [[TMP4]], [[TMP5]]
+// CHECK-NEXT:    store volatile <16 x i8> [[OR2]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[OR3:%.*]] = or <16 x i8> [[TMP6]], [[TMP7]]
+// CHECK-NEXT:    store volatile <16 x i8> [[OR3]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[OR4:%.*]] = or <16 x i8> [[TMP8]], [[TMP9]]
+// CHECK-NEXT:    store volatile <16 x i8> [[OR4]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[OR5:%.*]] = or <16 x i8> [[TMP10]], [[TMP11]]
+// CHECK-NEXT:    store volatile <16 x i8> [[OR5]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[OR6:%.*]] = or <16 x i8> [[TMP12]], [[TMP13]]
+// CHECK-NEXT:    store volatile <16 x i8> [[OR6]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[OR7:%.*]] = or <8 x i16> [[TMP14]], [[TMP15]]
+// CHECK-NEXT:    store volatile <8 x i16> [[OR7]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[OR8:%.*]] = or <8 x i16> [[TMP16]], [[TMP17]]
+// CHECK-NEXT:    store volatile <8 x i16> [[OR8]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[OR9:%.*]] = or <8 x i16> [[TMP18]], [[TMP19]]
+// CHECK-NEXT:    store volatile <8 x i16> [[OR9]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[OR10:%.*]] = or <8 x i16> [[TMP20]], [[TMP21]]
+// CHECK-NEXT:    store volatile <8 x i16> [[OR10]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[OR11:%.*]] = or <8 x i16> [[TMP22]], [[TMP23]]
+// CHECK-NEXT:    store volatile <8 x i16> [[OR11]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[OR12:%.*]] = or <8 x i16> [[TMP24]], [[TMP25]]
+// CHECK-NEXT:    store volatile <8 x i16> [[OR12]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP26:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP27:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[OR13:%.*]] = or <8 x i16> [[TMP26]], [[TMP27]]
+// CHECK-NEXT:    store volatile <8 x i16> [[OR13]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP28:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[OR14:%.*]] = or <4 x i32> [[TMP28]], [[TMP29]]
+// CHECK-NEXT:    store volatile <4 x i32> [[OR14]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP30:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP31:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[OR15:%.*]] = or <4 x i32> [[TMP30]], [[TMP31]]
+// CHECK-NEXT:    store volatile <4 x i32> [[OR15]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP32:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP33:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[OR16:%.*]] = or <4 x i32> [[TMP32]], [[TMP33]]
+// CHECK-NEXT:    store volatile <4 x i32> [[OR16]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP34:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP35:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[OR17:%.*]] = or <4 x i32> [[TMP34]], [[TMP35]]
+// CHECK-NEXT:    store volatile <4 x i32> [[OR17]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP36:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP37:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[OR18:%.*]] = or <4 x i32> [[TMP36]], [[TMP37]]
+// CHECK-NEXT:    store volatile <4 x i32> [[OR18]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP38:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP39:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[OR19:%.*]] = or <4 x i32> [[TMP38]], [[TMP39]]
+// CHECK-NEXT:    store volatile <4 x i32> [[OR19]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP40:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP41:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[OR20:%.*]] = or <4 x i32> [[TMP40]], [[TMP41]]
+// CHECK-NEXT:    store volatile <4 x i32> [[OR20]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP42:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP43:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[OR21:%.*]] = or <2 x i64> [[TMP42]], [[TMP43]]
+// CHECK-NEXT:    store volatile <2 x i64> [[OR21]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP44:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP45:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[OR22:%.*]] = or <2 x i64> [[TMP44]], [[TMP45]]
+// CHECK-NEXT:    store volatile <2 x i64> [[OR22]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP46:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP47:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[OR23:%.*]] = or <2 x i64> [[TMP46]], [[TMP47]]
+// CHECK-NEXT:    store volatile <2 x i64> [[OR23]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP48:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP49:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[OR24:%.*]] = or <2 x i64> [[TMP48]], [[TMP49]]
+// CHECK-NEXT:    store volatile <2 x i64> [[OR24]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP50:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP51:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[OR25:%.*]] = or <2 x i64> [[TMP50]], [[TMP51]]
+// CHECK-NEXT:    store volatile <2 x i64> [[OR25]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP52:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP53:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[OR26:%.*]] = or <2 x i64> [[TMP52]], [[TMP53]]
+// CHECK-NEXT:    store volatile <2 x i64> [[OR26]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP54:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP55:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[OR27:%.*]] = or <2 x i64> [[TMP54]], [[TMP55]]
+// CHECK-NEXT:    store volatile <2 x i64> [[OR27]], ptr @bl, align 8
+// CHECK-NEXT:    ret void
+//
 void test_or(void) {
 
   sc = sc | sc2;
@@ -1548,88 +1612,91 @@ void test_or(void) {
   bl = bl | bl2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_or_assign() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[OR:%.*]] = or <16 x i8> [[TMP1]], [[TMP0]]
-// CHECK:   store volatile <16 x i8> [[OR]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[OR1:%.*]] = or <16 x i8> [[TMP3]], [[TMP2]]
-// CHECK:   store volatile <16 x i8> [[OR1]], ptr @sc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[OR2:%.*]] = or <16 x i8> [[TMP5]], [[TMP4]]
-// CHECK:   store volatile <16 x i8> [[OR2]], ptr @uc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[OR3:%.*]] = or <16 x i8> [[TMP7]], [[TMP6]]
-// CHECK:   store volatile <16 x i8> [[OR3]], ptr @uc, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[OR4:%.*]] = or <16 x i8> [[TMP9]], [[TMP8]]
-// CHECK:   store volatile <16 x i8> [[OR4]], ptr @bc, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[OR5:%.*]] = or <8 x i16> [[TMP11]], [[TMP10]]
-// CHECK:   store volatile <8 x i16> [[OR5]], ptr @ss, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[OR6:%.*]] = or <8 x i16> [[TMP13]], [[TMP12]]
-// CHECK:   store volatile <8 x i16> [[OR6]], ptr @ss, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[OR7:%.*]] = or <8 x i16> [[TMP15]], [[TMP14]]
-// CHECK:   store volatile <8 x i16> [[OR7]], ptr @us, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[OR8:%.*]] = or <8 x i16> [[TMP17]], [[TMP16]]
-// CHECK:   store volatile <8 x i16> [[OR8]], ptr @us, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[TMP19:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[OR9:%.*]] = or <8 x i16> [[TMP19]], [[TMP18]]
-// CHECK:   store volatile <8 x i16> [[OR9]], ptr @bs, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[OR10:%.*]] = or <4 x i32> [[TMP21]], [[TMP20]]
-// CHECK:   store volatile <4 x i32> [[OR10]], ptr @si, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[OR11:%.*]] = or <4 x i32> [[TMP23]], [[TMP22]]
-// CHECK:   store volatile <4 x i32> [[OR11]], ptr @si, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[OR12:%.*]] = or <4 x i32> [[TMP25]], [[TMP24]]
-// CHECK:   store volatile <4 x i32> [[OR12]], ptr @ui, align 8
-// CHECK:   [[TMP26:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[TMP27:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[OR13:%.*]] = or <4 x i32> [[TMP27]], [[TMP26]]
-// CHECK:   store volatile <4 x i32> [[OR13]], ptr @ui, align 8
-// CHECK:   [[TMP28:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[TMP29:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[OR14:%.*]] = or <4 x i32> [[TMP29]], [[TMP28]]
-// CHECK:   store volatile <4 x i32> [[OR14]], ptr @bi, align 8
-// CHECK:   [[TMP30:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[TMP31:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[OR15:%.*]] = or <2 x i64> [[TMP31]], [[TMP30]]
-// CHECK:   store volatile <2 x i64> [[OR15]], ptr @sl, align 8
-// CHECK:   [[TMP32:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[TMP33:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[OR16:%.*]] = or <2 x i64> [[TMP33]], [[TMP32]]
-// CHECK:   store volatile <2 x i64> [[OR16]], ptr @sl, align 8
-// CHECK:   [[TMP34:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[TMP35:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[OR17:%.*]] = or <2 x i64> [[TMP35]], [[TMP34]]
-// CHECK:   store volatile <2 x i64> [[OR17]], ptr @ul, align 8
-// CHECK:   [[TMP36:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[TMP37:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[OR18:%.*]] = or <2 x i64> [[TMP37]], [[TMP36]]
-// CHECK:   store volatile <2 x i64> [[OR18]], ptr @ul, align 8
-// CHECK:   [[TMP38:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[TMP39:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[OR19:%.*]] = or <2 x i64> [[TMP39]], [[TMP38]]
-// CHECK:   store volatile <2 x i64> [[OR19]], ptr @bl, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_or_assign(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[OR:%.*]] = or <16 x i8> [[TMP1]], [[TMP0]]
+// CHECK-NEXT:    store volatile <16 x i8> [[OR]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[OR1:%.*]] = or <16 x i8> [[TMP3]], [[TMP2]]
+// CHECK-NEXT:    store volatile <16 x i8> [[OR1]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[OR2:%.*]] = or <16 x i8> [[TMP5]], [[TMP4]]
+// CHECK-NEXT:    store volatile <16 x i8> [[OR2]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[OR3:%.*]] = or <16 x i8> [[TMP7]], [[TMP6]]
+// CHECK-NEXT:    store volatile <16 x i8> [[OR3]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[OR4:%.*]] = or <16 x i8> [[TMP9]], [[TMP8]]
+// CHECK-NEXT:    store volatile <16 x i8> [[OR4]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[OR5:%.*]] = or <8 x i16> [[TMP11]], [[TMP10]]
+// CHECK-NEXT:    store volatile <8 x i16> [[OR5]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[OR6:%.*]] = or <8 x i16> [[TMP13]], [[TMP12]]
+// CHECK-NEXT:    store volatile <8 x i16> [[OR6]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[OR7:%.*]] = or <8 x i16> [[TMP15]], [[TMP14]]
+// CHECK-NEXT:    store volatile <8 x i16> [[OR7]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[OR8:%.*]] = or <8 x i16> [[TMP17]], [[TMP16]]
+// CHECK-NEXT:    store volatile <8 x i16> [[OR8]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[OR9:%.*]] = or <8 x i16> [[TMP19]], [[TMP18]]
+// CHECK-NEXT:    store volatile <8 x i16> [[OR9]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[OR10:%.*]] = or <4 x i32> [[TMP21]], [[TMP20]]
+// CHECK-NEXT:    store volatile <4 x i32> [[OR10]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[OR11:%.*]] = or <4 x i32> [[TMP23]], [[TMP22]]
+// CHECK-NEXT:    store volatile <4 x i32> [[OR11]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[OR12:%.*]] = or <4 x i32> [[TMP25]], [[TMP24]]
+// CHECK-NEXT:    store volatile <4 x i32> [[OR12]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP26:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[TMP27:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[OR13:%.*]] = or <4 x i32> [[TMP27]], [[TMP26]]
+// CHECK-NEXT:    store volatile <4 x i32> [[OR13]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP28:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[TMP29:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[OR14:%.*]] = or <4 x i32> [[TMP29]], [[TMP28]]
+// CHECK-NEXT:    store volatile <4 x i32> [[OR14]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP30:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[TMP31:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[OR15:%.*]] = or <2 x i64> [[TMP31]], [[TMP30]]
+// CHECK-NEXT:    store volatile <2 x i64> [[OR15]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP32:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[TMP33:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[OR16:%.*]] = or <2 x i64> [[TMP33]], [[TMP32]]
+// CHECK-NEXT:    store volatile <2 x i64> [[OR16]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP34:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[TMP35:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[OR17:%.*]] = or <2 x i64> [[TMP35]], [[TMP34]]
+// CHECK-NEXT:    store volatile <2 x i64> [[OR17]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP36:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[TMP37:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[OR18:%.*]] = or <2 x i64> [[TMP37]], [[TMP36]]
+// CHECK-NEXT:    store volatile <2 x i64> [[OR18]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP38:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[TMP39:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[OR19:%.*]] = or <2 x i64> [[TMP39]], [[TMP38]]
+// CHECK-NEXT:    store volatile <2 x i64> [[OR19]], ptr @bl, align 8
+// CHECK-NEXT:    ret void
+//
 void test_or_assign(void) {
 
   sc |= sc2;
@@ -1657,120 +1724,123 @@ void test_or_assign(void) {
   bl |= bl2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_xor() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[XOR:%.*]] = xor <16 x i8> [[TMP0]], [[TMP1]]
-// CHECK:   store volatile <16 x i8> [[XOR]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[XOR1:%.*]] = xor <16 x i8> [[TMP2]], [[TMP3]]
-// CHECK:   store volatile <16 x i8> [[XOR1]], ptr @sc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[XOR2:%.*]] = xor <16 x i8> [[TMP4]], [[TMP5]]
-// CHECK:   store volatile <16 x i8> [[XOR2]], ptr @sc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[XOR3:%.*]] = xor <16 x i8> [[TMP6]], [[TMP7]]
-// CHECK:   store volatile <16 x i8> [[XOR3]], ptr @uc, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[XOR4:%.*]] = xor <16 x i8> [[TMP8]], [[TMP9]]
-// CHECK:   store volatile <16 x i8> [[XOR4]], ptr @uc, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[XOR5:%.*]] = xor <16 x i8> [[TMP10]], [[TMP11]]
-// CHECK:   store volatile <16 x i8> [[XOR5]], ptr @uc, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[XOR6:%.*]] = xor <16 x i8> [[TMP12]], [[TMP13]]
-// CHECK:   store volatile <16 x i8> [[XOR6]], ptr @bc, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[XOR7:%.*]] = xor <8 x i16> [[TMP14]], [[TMP15]]
-// CHECK:   store volatile <8 x i16> [[XOR7]], ptr @ss, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[XOR8:%.*]] = xor <8 x i16> [[TMP16]], [[TMP17]]
-// CHECK:   store volatile <8 x i16> [[XOR8]], ptr @ss, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP19:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[XOR9:%.*]] = xor <8 x i16> [[TMP18]], [[TMP19]]
-// CHECK:   store volatile <8 x i16> [[XOR9]], ptr @ss, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[XOR10:%.*]] = xor <8 x i16> [[TMP20]], [[TMP21]]
-// CHECK:   store volatile <8 x i16> [[XOR10]], ptr @us, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[XOR11:%.*]] = xor <8 x i16> [[TMP22]], [[TMP23]]
-// CHECK:   store volatile <8 x i16> [[XOR11]], ptr @us, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[XOR12:%.*]] = xor <8 x i16> [[TMP24]], [[TMP25]]
-// CHECK:   store volatile <8 x i16> [[XOR12]], ptr @us, align 8
-// CHECK:   [[TMP26:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP27:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[XOR13:%.*]] = xor <8 x i16> [[TMP26]], [[TMP27]]
-// CHECK:   store volatile <8 x i16> [[XOR13]], ptr @bs, align 8
-// CHECK:   [[TMP28:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[XOR14:%.*]] = xor <4 x i32> [[TMP28]], [[TMP29]]
-// CHECK:   store volatile <4 x i32> [[XOR14]], ptr @si, align 8
-// CHECK:   [[TMP30:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP31:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[XOR15:%.*]] = xor <4 x i32> [[TMP30]], [[TMP31]]
-// CHECK:   store volatile <4 x i32> [[XOR15]], ptr @si, align 8
-// CHECK:   [[TMP32:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP33:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[XOR16:%.*]] = xor <4 x i32> [[TMP32]], [[TMP33]]
-// CHECK:   store volatile <4 x i32> [[XOR16]], ptr @si, align 8
-// CHECK:   [[TMP34:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP35:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[XOR17:%.*]] = xor <4 x i32> [[TMP34]], [[TMP35]]
-// CHECK:   store volatile <4 x i32> [[XOR17]], ptr @ui, align 8
-// CHECK:   [[TMP36:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP37:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[XOR18:%.*]] = xor <4 x i32> [[TMP36]], [[TMP37]]
-// CHECK:   store volatile <4 x i32> [[XOR18]], ptr @ui, align 8
-// CHECK:   [[TMP38:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP39:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[XOR19:%.*]] = xor <4 x i32> [[TMP38]], [[TMP39]]
-// CHECK:   store volatile <4 x i32> [[XOR19]], ptr @ui, align 8
-// CHECK:   [[TMP40:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP41:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[XOR20:%.*]] = xor <4 x i32> [[TMP40]], [[TMP41]]
-// CHECK:   store volatile <4 x i32> [[XOR20]], ptr @bi, align 8
-// CHECK:   [[TMP42:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP43:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[XOR21:%.*]] = xor <2 x i64> [[TMP42]], [[TMP43]]
-// CHECK:   store volatile <2 x i64> [[XOR21]], ptr @sl, align 8
-// CHECK:   [[TMP44:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP45:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[XOR22:%.*]] = xor <2 x i64> [[TMP44]], [[TMP45]]
-// CHECK:   store volatile <2 x i64> [[XOR22]], ptr @sl, align 8
-// CHECK:   [[TMP46:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP47:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[XOR23:%.*]] = xor <2 x i64> [[TMP46]], [[TMP47]]
-// CHECK:   store volatile <2 x i64> [[XOR23]], ptr @sl, align 8
-// CHECK:   [[TMP48:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP49:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[XOR24:%.*]] = xor <2 x i64> [[TMP48]], [[TMP49]]
-// CHECK:   store volatile <2 x i64> [[XOR24]], ptr @ul, align 8
-// CHECK:   [[TMP50:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP51:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[XOR25:%.*]] = xor <2 x i64> [[TMP50]], [[TMP51]]
-// CHECK:   store volatile <2 x i64> [[XOR25]], ptr @ul, align 8
-// CHECK:   [[TMP52:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP53:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[XOR26:%.*]] = xor <2 x i64> [[TMP52]], [[TMP53]]
-// CHECK:   store volatile <2 x i64> [[XOR26]], ptr @ul, align 8
-// CHECK:   [[TMP54:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP55:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[XOR27:%.*]] = xor <2 x i64> [[TMP54]], [[TMP55]]
-// CHECK:   store volatile <2 x i64> [[XOR27]], ptr @bl, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_xor(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[XOR:%.*]] = xor <16 x i8> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    store volatile <16 x i8> [[XOR]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[XOR1:%.*]] = xor <16 x i8> [[TMP2]], [[TMP3]]
+// CHECK-NEXT:    store volatile <16 x i8> [[XOR1]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[XOR2:%.*]] = xor <16 x i8> [[TMP4]], [[TMP5]]
+// CHECK-NEXT:    store volatile <16 x i8> [[XOR2]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[XOR3:%.*]] = xor <16 x i8> [[TMP6]], [[TMP7]]
+// CHECK-NEXT:    store volatile <16 x i8> [[XOR3]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[XOR4:%.*]] = xor <16 x i8> [[TMP8]], [[TMP9]]
+// CHECK-NEXT:    store volatile <16 x i8> [[XOR4]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[XOR5:%.*]] = xor <16 x i8> [[TMP10]], [[TMP11]]
+// CHECK-NEXT:    store volatile <16 x i8> [[XOR5]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[XOR6:%.*]] = xor <16 x i8> [[TMP12]], [[TMP13]]
+// CHECK-NEXT:    store volatile <16 x i8> [[XOR6]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[XOR7:%.*]] = xor <8 x i16> [[TMP14]], [[TMP15]]
+// CHECK-NEXT:    store volatile <8 x i16> [[XOR7]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[XOR8:%.*]] = xor <8 x i16> [[TMP16]], [[TMP17]]
+// CHECK-NEXT:    store volatile <8 x i16> [[XOR8]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[XOR9:%.*]] = xor <8 x i16> [[TMP18]], [[TMP19]]
+// CHECK-NEXT:    store volatile <8 x i16> [[XOR9]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[XOR10:%.*]] = xor <8 x i16> [[TMP20]], [[TMP21]]
+// CHECK-NEXT:    store volatile <8 x i16> [[XOR10]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[XOR11:%.*]] = xor <8 x i16> [[TMP22]], [[TMP23]]
+// CHECK-NEXT:    store volatile <8 x i16> [[XOR11]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[XOR12:%.*]] = xor <8 x i16> [[TMP24]], [[TMP25]]
+// CHECK-NEXT:    store volatile <8 x i16> [[XOR12]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP26:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP27:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[XOR13:%.*]] = xor <8 x i16> [[TMP26]], [[TMP27]]
+// CHECK-NEXT:    store volatile <8 x i16> [[XOR13]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP28:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[XOR14:%.*]] = xor <4 x i32> [[TMP28]], [[TMP29]]
+// CHECK-NEXT:    store volatile <4 x i32> [[XOR14]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP30:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP31:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[XOR15:%.*]] = xor <4 x i32> [[TMP30]], [[TMP31]]
+// CHECK-NEXT:    store volatile <4 x i32> [[XOR15]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP32:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP33:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[XOR16:%.*]] = xor <4 x i32> [[TMP32]], [[TMP33]]
+// CHECK-NEXT:    store volatile <4 x i32> [[XOR16]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP34:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP35:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[XOR17:%.*]] = xor <4 x i32> [[TMP34]], [[TMP35]]
+// CHECK-NEXT:    store volatile <4 x i32> [[XOR17]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP36:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP37:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[XOR18:%.*]] = xor <4 x i32> [[TMP36]], [[TMP37]]
+// CHECK-NEXT:    store volatile <4 x i32> [[XOR18]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP38:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP39:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[XOR19:%.*]] = xor <4 x i32> [[TMP38]], [[TMP39]]
+// CHECK-NEXT:    store volatile <4 x i32> [[XOR19]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP40:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP41:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[XOR20:%.*]] = xor <4 x i32> [[TMP40]], [[TMP41]]
+// CHECK-NEXT:    store volatile <4 x i32> [[XOR20]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP42:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP43:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[XOR21:%.*]] = xor <2 x i64> [[TMP42]], [[TMP43]]
+// CHECK-NEXT:    store volatile <2 x i64> [[XOR21]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP44:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP45:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[XOR22:%.*]] = xor <2 x i64> [[TMP44]], [[TMP45]]
+// CHECK-NEXT:    store volatile <2 x i64> [[XOR22]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP46:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP47:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[XOR23:%.*]] = xor <2 x i64> [[TMP46]], [[TMP47]]
+// CHECK-NEXT:    store volatile <2 x i64> [[XOR23]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP48:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP49:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[XOR24:%.*]] = xor <2 x i64> [[TMP48]], [[TMP49]]
+// CHECK-NEXT:    store volatile <2 x i64> [[XOR24]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP50:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP51:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[XOR25:%.*]] = xor <2 x i64> [[TMP50]], [[TMP51]]
+// CHECK-NEXT:    store volatile <2 x i64> [[XOR25]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP52:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP53:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[XOR26:%.*]] = xor <2 x i64> [[TMP52]], [[TMP53]]
+// CHECK-NEXT:    store volatile <2 x i64> [[XOR26]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP54:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP55:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[XOR27:%.*]] = xor <2 x i64> [[TMP54]], [[TMP55]]
+// CHECK-NEXT:    store volatile <2 x i64> [[XOR27]], ptr @bl, align 8
+// CHECK-NEXT:    ret void
+//
 void test_xor(void) {
 
   sc = sc ^ sc2;
@@ -1806,88 +1876,91 @@ void test_xor(void) {
   bl = bl ^ bl2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_xor_assign() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[XOR:%.*]] = xor <16 x i8> [[TMP1]], [[TMP0]]
-// CHECK:   store volatile <16 x i8> [[XOR]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[XOR1:%.*]] = xor <16 x i8> [[TMP3]], [[TMP2]]
-// CHECK:   store volatile <16 x i8> [[XOR1]], ptr @sc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[XOR2:%.*]] = xor <16 x i8> [[TMP5]], [[TMP4]]
-// CHECK:   store volatile <16 x i8> [[XOR2]], ptr @uc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[XOR3:%.*]] = xor <16 x i8> [[TMP7]], [[TMP6]]
-// CHECK:   store volatile <16 x i8> [[XOR3]], ptr @uc, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[XOR4:%.*]] = xor <16 x i8> [[TMP9]], [[TMP8]]
-// CHECK:   store volatile <16 x i8> [[XOR4]], ptr @bc, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[XOR5:%.*]] = xor <8 x i16> [[TMP11]], [[TMP10]]
-// CHECK:   store volatile <8 x i16> [[XOR5]], ptr @ss, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[XOR6:%.*]] = xor <8 x i16> [[TMP13]], [[TMP12]]
-// CHECK:   store volatile <8 x i16> [[XOR6]], ptr @ss, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[XOR7:%.*]] = xor <8 x i16> [[TMP15]], [[TMP14]]
-// CHECK:   store volatile <8 x i16> [[XOR7]], ptr @us, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[XOR8:%.*]] = xor <8 x i16> [[TMP17]], [[TMP16]]
-// CHECK:   store volatile <8 x i16> [[XOR8]], ptr @us, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[TMP19:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[XOR9:%.*]] = xor <8 x i16> [[TMP19]], [[TMP18]]
-// CHECK:   store volatile <8 x i16> [[XOR9]], ptr @bs, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[XOR10:%.*]] = xor <4 x i32> [[TMP21]], [[TMP20]]
-// CHECK:   store volatile <4 x i32> [[XOR10]], ptr @si, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[XOR11:%.*]] = xor <4 x i32> [[TMP23]], [[TMP22]]
-// CHECK:   store volatile <4 x i32> [[XOR11]], ptr @si, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[XOR12:%.*]] = xor <4 x i32> [[TMP25]], [[TMP24]]
-// CHECK:   store volatile <4 x i32> [[XOR12]], ptr @ui, align 8
-// CHECK:   [[TMP26:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[TMP27:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[XOR13:%.*]] = xor <4 x i32> [[TMP27]], [[TMP26]]
-// CHECK:   store volatile <4 x i32> [[XOR13]], ptr @ui, align 8
-// CHECK:   [[TMP28:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[TMP29:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[XOR14:%.*]] = xor <4 x i32> [[TMP29]], [[TMP28]]
-// CHECK:   store volatile <4 x i32> [[XOR14]], ptr @bi, align 8
-// CHECK:   [[TMP30:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[TMP31:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[XOR15:%.*]] = xor <2 x i64> [[TMP31]], [[TMP30]]
-// CHECK:   store volatile <2 x i64> [[XOR15]], ptr @sl, align 8
-// CHECK:   [[TMP32:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[TMP33:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[XOR16:%.*]] = xor <2 x i64> [[TMP33]], [[TMP32]]
-// CHECK:   store volatile <2 x i64> [[XOR16]], ptr @sl, align 8
-// CHECK:   [[TMP34:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[TMP35:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[XOR17:%.*]] = xor <2 x i64> [[TMP35]], [[TMP34]]
-// CHECK:   store volatile <2 x i64> [[XOR17]], ptr @ul, align 8
-// CHECK:   [[TMP36:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[TMP37:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[XOR18:%.*]] = xor <2 x i64> [[TMP37]], [[TMP36]]
-// CHECK:   store volatile <2 x i64> [[XOR18]], ptr @ul, align 8
-// CHECK:   [[TMP38:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[TMP39:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[XOR19:%.*]] = xor <2 x i64> [[TMP39]], [[TMP38]]
-// CHECK:   store volatile <2 x i64> [[XOR19]], ptr @bl, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_xor_assign(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[XOR:%.*]] = xor <16 x i8> [[TMP1]], [[TMP0]]
+// CHECK-NEXT:    store volatile <16 x i8> [[XOR]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[XOR1:%.*]] = xor <16 x i8> [[TMP3]], [[TMP2]]
+// CHECK-NEXT:    store volatile <16 x i8> [[XOR1]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[XOR2:%.*]] = xor <16 x i8> [[TMP5]], [[TMP4]]
+// CHECK-NEXT:    store volatile <16 x i8> [[XOR2]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[XOR3:%.*]] = xor <16 x i8> [[TMP7]], [[TMP6]]
+// CHECK-NEXT:    store volatile <16 x i8> [[XOR3]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[XOR4:%.*]] = xor <16 x i8> [[TMP9]], [[TMP8]]
+// CHECK-NEXT:    store volatile <16 x i8> [[XOR4]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[XOR5:%.*]] = xor <8 x i16> [[TMP11]], [[TMP10]]
+// CHECK-NEXT:    store volatile <8 x i16> [[XOR5]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[XOR6:%.*]] = xor <8 x i16> [[TMP13]], [[TMP12]]
+// CHECK-NEXT:    store volatile <8 x i16> [[XOR6]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[XOR7:%.*]] = xor <8 x i16> [[TMP15]], [[TMP14]]
+// CHECK-NEXT:    store volatile <8 x i16> [[XOR7]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[XOR8:%.*]] = xor <8 x i16> [[TMP17]], [[TMP16]]
+// CHECK-NEXT:    store volatile <8 x i16> [[XOR8]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[XOR9:%.*]] = xor <8 x i16> [[TMP19]], [[TMP18]]
+// CHECK-NEXT:    store volatile <8 x i16> [[XOR9]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[XOR10:%.*]] = xor <4 x i32> [[TMP21]], [[TMP20]]
+// CHECK-NEXT:    store volatile <4 x i32> [[XOR10]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[XOR11:%.*]] = xor <4 x i32> [[TMP23]], [[TMP22]]
+// CHECK-NEXT:    store volatile <4 x i32> [[XOR11]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[XOR12:%.*]] = xor <4 x i32> [[TMP25]], [[TMP24]]
+// CHECK-NEXT:    store volatile <4 x i32> [[XOR12]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP26:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[TMP27:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[XOR13:%.*]] = xor <4 x i32> [[TMP27]], [[TMP26]]
+// CHECK-NEXT:    store volatile <4 x i32> [[XOR13]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP28:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[TMP29:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[XOR14:%.*]] = xor <4 x i32> [[TMP29]], [[TMP28]]
+// CHECK-NEXT:    store volatile <4 x i32> [[XOR14]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP30:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[TMP31:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[XOR15:%.*]] = xor <2 x i64> [[TMP31]], [[TMP30]]
+// CHECK-NEXT:    store volatile <2 x i64> [[XOR15]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP32:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[TMP33:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[XOR16:%.*]] = xor <2 x i64> [[TMP33]], [[TMP32]]
+// CHECK-NEXT:    store volatile <2 x i64> [[XOR16]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP34:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[TMP35:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[XOR17:%.*]] = xor <2 x i64> [[TMP35]], [[TMP34]]
+// CHECK-NEXT:    store volatile <2 x i64> [[XOR17]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP36:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[TMP37:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[XOR18:%.*]] = xor <2 x i64> [[TMP37]], [[TMP36]]
+// CHECK-NEXT:    store volatile <2 x i64> [[XOR18]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP38:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[TMP39:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[XOR19:%.*]] = xor <2 x i64> [[TMP39]], [[TMP38]]
+// CHECK-NEXT:    store volatile <2 x i64> [[XOR19]], ptr @bl, align 8
+// CHECK-NEXT:    ret void
+//
 void test_xor_assign(void) {
 
   sc ^= sc2;
@@ -1915,150 +1988,153 @@ void test_xor_assign(void) {
   bl ^= bl2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_sl() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[SHL:%.*]] = shl <16 x i8> [[TMP0]], [[TMP1]]
-// CHECK:   store volatile <16 x i8> [[SHL]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[SHL1:%.*]] = shl <16 x i8> [[TMP2]], [[TMP3]]
-// CHECK:   store volatile <16 x i8> [[SHL1]], ptr @sc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT:%.*]] = insertelement <16 x i32> poison, i32 [[TMP5]], i64 0
-// CHECK:   [[SPLAT_SPLAT:%.*]] = shufflevector <16 x i32> [[SPLAT_SPLAT:%.*]]insert, <16 x i32> poison, <16 x i32> zeroinitializer
-// CHECK:   [[SH_PROM:%.*]] = trunc <16 x i32> [[SPLAT_SPLAT]] to <16 x i8>
-// CHECK:   [[SHL2:%.*]] = shl <16 x i8> [[TMP4]], [[SH_PROM]]
-// CHECK:   store volatile <16 x i8> [[SHL2]], ptr @sc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[SHL3:%.*]] = shl <16 x i8> [[TMP6]], splat (i8 5)
-// CHECK:   store volatile <16 x i8> [[SHL3]], ptr @sc, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[SHL4:%.*]] = shl <16 x i8> [[TMP7]], [[TMP8]]
-// CHECK:   store volatile <16 x i8> [[SHL4]], ptr @uc, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[SHL5:%.*]] = shl <16 x i8> [[TMP9]], [[TMP10]]
-// CHECK:   store volatile <16 x i8> [[SHL5]], ptr @uc, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT6:%.*]] = insertelement <16 x i32> poison, i32 [[TMP12]], i64 0
-// CHECK:   [[SPLAT_SPLAT7:%.*]] = shufflevector <16 x i32> [[SPLAT_SPLATINSERT6]], <16 x i32> poison, <16 x i32> zeroinitializer
-// CHECK:   [[SH_PROM8:%.*]] = trunc <16 x i32> [[SPLAT_SPLAT7]] to <16 x i8>
-// CHECK:   [[SHL9:%.*]] = shl <16 x i8> [[TMP11]], [[SH_PROM8]]
-// CHECK:   store volatile <16 x i8> [[SHL9]], ptr @uc, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[SHL10:%.*]] = shl <16 x i8> [[TMP13]], splat (i8 5)
-// CHECK:   store volatile <16 x i8> [[SHL10]], ptr @uc, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[SHL11:%.*]] = shl <8 x i16> [[TMP14]], [[TMP15]]
-// CHECK:   store volatile <8 x i16> [[SHL11]], ptr @ss, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[SHL12:%.*]] = shl <8 x i16> [[TMP16]], [[TMP17]]
-// CHECK:   store volatile <8 x i16> [[SHL12]], ptr @ss, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP19:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT13:%.*]] = insertelement <8 x i32> poison, i32 [[TMP19]], i64 0
-// CHECK:   [[SPLAT_SPLAT14:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT13]], <8 x i32> poison, <8 x i32> zeroinitializer
-// CHECK:   [[SH_PROM15:%.*]] = trunc <8 x i32> [[SPLAT_SPLAT14]] to <8 x i16>
-// CHECK:   [[SHL16:%.*]] = shl <8 x i16> [[TMP18]], [[SH_PROM15]]
-// CHECK:   store volatile <8 x i16> [[SHL16]], ptr @ss, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[SHL17:%.*]] = shl <8 x i16> [[TMP20]], splat (i16 5)
-// CHECK:   store volatile <8 x i16> [[SHL17]], ptr @ss, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[SHL18:%.*]] = shl <8 x i16> [[TMP21]], [[TMP22]]
-// CHECK:   store volatile <8 x i16> [[SHL18]], ptr @us, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[SHL19:%.*]] = shl <8 x i16> [[TMP23]], [[TMP24]]
-// CHECK:   store volatile <8 x i16> [[SHL19]], ptr @us, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP26:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT20:%.*]] = insertelement <8 x i32> poison, i32 [[TMP26]], i64 0
-// CHECK:   [[SPLAT_SPLAT21:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT20]], <8 x i32> poison, <8 x i32> zeroinitializer
-// CHECK:   [[SH_PROM22:%.*]] = trunc <8 x i32> [[SPLAT_SPLAT21]] to <8 x i16>
-// CHECK:   [[SHL23:%.*]] = shl <8 x i16> [[TMP25]], [[SH_PROM22]]
-// CHECK:   store volatile <8 x i16> [[SHL23]], ptr @us, align 8
-// CHECK:   [[TMP27:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[SHL24:%.*]] = shl <8 x i16> [[TMP27]], splat (i16 5)
-// CHECK:   store volatile <8 x i16> [[SHL24]], ptr @us, align 8
-// CHECK:   [[TMP28:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[SHL25:%.*]] = shl <4 x i32> [[TMP28]], [[TMP29]]
-// CHECK:   store volatile <4 x i32> [[SHL25]], ptr @si, align 8
-// CHECK:   [[TMP30:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP31:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[SHL26:%.*]] = shl <4 x i32> [[TMP30]], [[TMP31]]
-// CHECK:   store volatile <4 x i32> [[SHL26]], ptr @si, align 8
-// CHECK:   [[TMP32:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP33:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT27:%.*]] = insertelement <4 x i32> poison, i32 [[TMP33]], i64 0
-// CHECK:   [[SPLAT_SPLAT28:%.*]] = shufflevector <4 x i32> [[SPLAT_SPLATINSERT27]], <4 x i32> poison, <4 x i32> zeroinitializer
-// CHECK:   [[SHL29:%.*]] = shl <4 x i32> [[TMP32]], [[SPLAT_SPLAT28]]
-// CHECK:   store volatile <4 x i32> [[SHL29]], ptr @si, align 8
-// CHECK:   [[TMP34:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[SHL30:%.*]] = shl <4 x i32> [[TMP34]], splat (i32 5)
-// CHECK:   store volatile <4 x i32> [[SHL30]], ptr @si, align 8
-// CHECK:   [[TMP35:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP36:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[SHL31:%.*]] = shl <4 x i32> [[TMP35]], [[TMP36]]
-// CHECK:   store volatile <4 x i32> [[SHL31]], ptr @ui, align 8
-// CHECK:   [[TMP37:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP38:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[SHL32:%.*]] = shl <4 x i32> [[TMP37]], [[TMP38]]
-// CHECK:   store volatile <4 x i32> [[SHL32]], ptr @ui, align 8
-// CHECK:   [[TMP39:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP40:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT33:%.*]] = insertelement <4 x i32> poison, i32 [[TMP40]], i64 0
-// CHECK:   [[SPLAT_SPLAT34:%.*]] = shufflevector <4 x i32> [[SPLAT_SPLATINSERT33]], <4 x i32> poison, <4 x i32> zeroinitializer
-// CHECK:   [[SHL35:%.*]] = shl <4 x i32> [[TMP39]], [[SPLAT_SPLAT34]]
-// CHECK:   store volatile <4 x i32> [[SHL35]], ptr @ui, align 8
-// CHECK:   [[TMP41:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[SHL36:%.*]] = shl <4 x i32> [[TMP41]], splat (i32 5)
-// CHECK:   store volatile <4 x i32> [[SHL36]], ptr @ui, align 8
-// CHECK:   [[TMP42:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP43:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[SHL37:%.*]] = shl <2 x i64> [[TMP42]], [[TMP43]]
-// CHECK:   store volatile <2 x i64> [[SHL37]], ptr @sl, align 8
-// CHECK:   [[TMP44:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP45:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[SHL38:%.*]] = shl <2 x i64> [[TMP44]], [[TMP45]]
-// CHECK:   store volatile <2 x i64> [[SHL38]], ptr @sl, align 8
-// CHECK:   [[TMP46:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP47:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT39:%.*]] = insertelement <2 x i32> poison, i32 [[TMP47]], i64 0
-// CHECK:   [[SPLAT_SPLAT40:%.*]] = shufflevector <2 x i32> [[SPLAT_SPLATINSERT39]], <2 x i32> poison, <2 x i32> zeroinitializer
-// CHECK:   [[SH_PROM41:%.*]] = zext <2 x i32> [[SPLAT_SPLAT40]] to <2 x i64>
-// CHECK:   [[SHL42:%.*]] = shl <2 x i64> [[TMP46]], [[SH_PROM41]]
-// CHECK:   store volatile <2 x i64> [[SHL42]], ptr @sl, align 8
-// CHECK:   [[TMP48:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[SHL43:%.*]] = shl <2 x i64> [[TMP48]], splat (i64 5)
-// CHECK:   store volatile <2 x i64> [[SHL43]], ptr @sl, align 8
-// CHECK:   [[TMP49:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP50:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[SHL44:%.*]] = shl <2 x i64> [[TMP49]], [[TMP50]]
-// CHECK:   store volatile <2 x i64> [[SHL44]], ptr @ul, align 8
-// CHECK:   [[TMP51:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP52:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[SHL45:%.*]] = shl <2 x i64> [[TMP51]], [[TMP52]]
-// CHECK:   store volatile <2 x i64> [[SHL45]], ptr @ul, align 8
-// CHECK:   [[TMP53:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP54:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT46:%.*]] = insertelement <2 x i32> poison, i32 [[TMP54]], i64 0
-// CHECK:   [[SPLAT_SPLAT47:%.*]] = shufflevector <2 x i32> [[SPLAT_SPLATINSERT46]], <2 x i32> poison, <2 x i32> zeroinitializer
-// CHECK:   [[SH_PROM48:%.*]] = zext <2 x i32> [[SPLAT_SPLAT47]] to <2 x i64>
-// CHECK:   [[SHL49:%.*]] = shl <2 x i64> [[TMP53]], [[SH_PROM48]]
-// CHECK:   store volatile <2 x i64> [[SHL49]], ptr @ul, align 8
-// CHECK:   [[TMP55:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[SHL50:%.*]] = shl <2 x i64> [[TMP55]], splat (i64 5)
-// CHECK:   store volatile <2 x i64> [[SHL50]], ptr @ul, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_sl(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[SHL:%.*]] = shl <16 x i8> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHL]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[SHL1:%.*]] = shl <16 x i8> [[TMP2]], [[TMP3]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHL1]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <16 x i32> poison, i32 [[TMP5]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <16 x i32> [[SPLAT_SPLATINSERT]], <16 x i32> poison, <16 x i32> zeroinitializer
+// CHECK-NEXT:    [[SH_PROM:%.*]] = trunc <16 x i32> [[SPLAT_SPLAT]] to <16 x i8>
+// CHECK-NEXT:    [[SHL2:%.*]] = shl <16 x i8> [[TMP4]], [[SH_PROM]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHL2]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[SHL3:%.*]] = shl <16 x i8> [[TMP6]], splat (i8 5)
+// CHECK-NEXT:    store volatile <16 x i8> [[SHL3]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[SHL4:%.*]] = shl <16 x i8> [[TMP7]], [[TMP8]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHL4]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[SHL5:%.*]] = shl <16 x i8> [[TMP9]], [[TMP10]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHL5]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT6:%.*]] = insertelement <16 x i32> poison, i32 [[TMP12]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT7:%.*]] = shufflevector <16 x i32> [[SPLAT_SPLATINSERT6]], <16 x i32> poison, <16 x i32> zeroinitializer
+// CHECK-NEXT:    [[SH_PROM8:%.*]] = trunc <16 x i32> [[SPLAT_SPLAT7]] to <16 x i8>
+// CHECK-NEXT:    [[SHL9:%.*]] = shl <16 x i8> [[TMP11]], [[SH_PROM8]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHL9]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[SHL10:%.*]] = shl <16 x i8> [[TMP13]], splat (i8 5)
+// CHECK-NEXT:    store volatile <16 x i8> [[SHL10]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[SHL11:%.*]] = shl <8 x i16> [[TMP14]], [[TMP15]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHL11]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[SHL12:%.*]] = shl <8 x i16> [[TMP16]], [[TMP17]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHL12]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT13:%.*]] = insertelement <8 x i32> poison, i32 [[TMP19]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT14:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT13]], <8 x i32> poison, <8 x i32> zeroinitializer
+// CHECK-NEXT:    [[SH_PROM15:%.*]] = trunc <8 x i32> [[SPLAT_SPLAT14]] to <8 x i16>
+// CHECK-NEXT:    [[SHL16:%.*]] = shl <8 x i16> [[TMP18]], [[SH_PROM15]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHL16]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[SHL17:%.*]] = shl <8 x i16> [[TMP20]], splat (i16 5)
+// CHECK-NEXT:    store volatile <8 x i16> [[SHL17]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[SHL18:%.*]] = shl <8 x i16> [[TMP21]], [[TMP22]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHL18]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[SHL19:%.*]] = shl <8 x i16> [[TMP23]], [[TMP24]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHL19]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP26:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT20:%.*]] = insertelement <8 x i32> poison, i32 [[TMP26]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT21:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT20]], <8 x i32> poison, <8 x i32> zeroinitializer
+// CHECK-NEXT:    [[SH_PROM22:%.*]] = trunc <8 x i32> [[SPLAT_SPLAT21]] to <8 x i16>
+// CHECK-NEXT:    [[SHL23:%.*]] = shl <8 x i16> [[TMP25]], [[SH_PROM22]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHL23]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP27:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[SHL24:%.*]] = shl <8 x i16> [[TMP27]], splat (i16 5)
+// CHECK-NEXT:    store volatile <8 x i16> [[SHL24]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP28:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[SHL25:%.*]] = shl <4 x i32> [[TMP28]], [[TMP29]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHL25]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP30:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP31:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[SHL26:%.*]] = shl <4 x i32> [[TMP30]], [[TMP31]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHL26]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP32:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP33:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT27:%.*]] = insertelement <4 x i32> poison, i32 [[TMP33]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT28:%.*]] = shufflevector <4 x i32> [[SPLAT_SPLATINSERT27]], <4 x i32> poison, <4 x i32> zeroinitializer
+// CHECK-NEXT:    [[SHL29:%.*]] = shl <4 x i32> [[TMP32]], [[SPLAT_SPLAT28]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHL29]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP34:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[SHL30:%.*]] = shl <4 x i32> [[TMP34]], splat (i32 5)
+// CHECK-NEXT:    store volatile <4 x i32> [[SHL30]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP35:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP36:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[SHL31:%.*]] = shl <4 x i32> [[TMP35]], [[TMP36]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHL31]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP37:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP38:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[SHL32:%.*]] = shl <4 x i32> [[TMP37]], [[TMP38]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHL32]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP39:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP40:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT33:%.*]] = insertelement <4 x i32> poison, i32 [[TMP40]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT34:%.*]] = shufflevector <4 x i32> [[SPLAT_SPLATINSERT33]], <4 x i32> poison, <4 x i32> zeroinitializer
+// CHECK-NEXT:    [[SHL35:%.*]] = shl <4 x i32> [[TMP39]], [[SPLAT_SPLAT34]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHL35]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP41:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[SHL36:%.*]] = shl <4 x i32> [[TMP41]], splat (i32 5)
+// CHECK-NEXT:    store volatile <4 x i32> [[SHL36]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP42:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP43:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[SHL37:%.*]] = shl <2 x i64> [[TMP42]], [[TMP43]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHL37]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP44:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP45:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[SHL38:%.*]] = shl <2 x i64> [[TMP44]], [[TMP45]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHL38]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP46:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP47:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT39:%.*]] = insertelement <2 x i32> poison, i32 [[TMP47]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT40:%.*]] = shufflevector <2 x i32> [[SPLAT_SPLATINSERT39]], <2 x i32> poison, <2 x i32> zeroinitializer
+// CHECK-NEXT:    [[SH_PROM41:%.*]] = zext <2 x i32> [[SPLAT_SPLAT40]] to <2 x i64>
+// CHECK-NEXT:    [[SHL42:%.*]] = shl <2 x i64> [[TMP46]], [[SH_PROM41]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHL42]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP48:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[SHL43:%.*]] = shl <2 x i64> [[TMP48]], splat (i64 5)
+// CHECK-NEXT:    store volatile <2 x i64> [[SHL43]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP49:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP50:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[SHL44:%.*]] = shl <2 x i64> [[TMP49]], [[TMP50]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHL44]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP51:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP52:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[SHL45:%.*]] = shl <2 x i64> [[TMP51]], [[TMP52]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHL45]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP53:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP54:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT46:%.*]] = insertelement <2 x i32> poison, i32 [[TMP54]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT47:%.*]] = shufflevector <2 x i32> [[SPLAT_SPLATINSERT46]], <2 x i32> poison, <2 x i32> zeroinitializer
+// CHECK-NEXT:    [[SH_PROM48:%.*]] = zext <2 x i32> [[SPLAT_SPLAT47]] to <2 x i64>
+// CHECK-NEXT:    [[SHL49:%.*]] = shl <2 x i64> [[TMP53]], [[SH_PROM48]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHL49]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP55:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[SHL50:%.*]] = shl <2 x i64> [[TMP55]], splat (i64 5)
+// CHECK-NEXT:    store volatile <2 x i64> [[SHL50]], ptr @ul, align 8
+// CHECK-NEXT:    ret void
+//
 void test_sl(void) {
 
   sc = sc << sc2;
@@ -2098,150 +2174,153 @@ void test_sl(void) {
   ul = ul << 5;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_sl_assign() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[SHL:%.*]] = shl <16 x i8> [[TMP1]], [[TMP0]]
-// CHECK:   store volatile <16 x i8> [[SHL]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[SHL1:%.*]] = shl <16 x i8> [[TMP3]], [[TMP2]]
-// CHECK:   store volatile <16 x i8> [[SHL1]], ptr @sc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT:%.*]] = insertelement <16 x i32> poison, i32 [[TMP4]], i64 0
-// CHECK:   [[SPLAT_SPLAT:%.*]] = shufflevector <16 x i32> [[SPLAT_SPLAT:%.*]]insert, <16 x i32> poison, <16 x i32> zeroinitializer
-// CHECK:   [[TMP5:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[SH_PROM:%.*]] = trunc <16 x i32> [[SPLAT_SPLAT]] to <16 x i8>
-// CHECK:   [[SHL2:%.*]] = shl <16 x i8> [[TMP5]], [[SH_PROM]]
-// CHECK:   store volatile <16 x i8> [[SHL2]], ptr @sc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[SHL3:%.*]] = shl <16 x i8> [[TMP6]], splat (i8 5)
-// CHECK:   store volatile <16 x i8> [[SHL3]], ptr @sc, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[SHL4:%.*]] = shl <16 x i8> [[TMP8]], [[TMP7]]
-// CHECK:   store volatile <16 x i8> [[SHL4]], ptr @uc, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[SHL5:%.*]] = shl <16 x i8> [[TMP10]], [[TMP9]]
-// CHECK:   store volatile <16 x i8> [[SHL5]], ptr @uc, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT6:%.*]] = insertelement <16 x i32> poison, i32 [[TMP11]], i64 0
-// CHECK:   [[SPLAT_SPLAT7:%.*]] = shufflevector <16 x i32> [[SPLAT_SPLATINSERT6]], <16 x i32> poison, <16 x i32> zeroinitializer
-// CHECK:   [[TMP12:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[SH_PROM8:%.*]] = trunc <16 x i32> [[SPLAT_SPLAT7]] to <16 x i8>
-// CHECK:   [[SHL9:%.*]] = shl <16 x i8> [[TMP12]], [[SH_PROM8]]
-// CHECK:   store volatile <16 x i8> [[SHL9]], ptr @uc, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[SHL10:%.*]] = shl <16 x i8> [[TMP13]], splat (i8 5)
-// CHECK:   store volatile <16 x i8> [[SHL10]], ptr @uc, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[SHL11:%.*]] = shl <8 x i16> [[TMP15]], [[TMP14]]
-// CHECK:   store volatile <8 x i16> [[SHL11]], ptr @ss, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[SHL12:%.*]] = shl <8 x i16> [[TMP17]], [[TMP16]]
-// CHECK:   store volatile <8 x i16> [[SHL12]], ptr @ss, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT13:%.*]] = insertelement <8 x i32> poison, i32 [[TMP18]], i64 0
-// CHECK:   [[SPLAT_SPLAT14:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT13]], <8 x i32> poison, <8 x i32> zeroinitializer
-// CHECK:   [[TMP19:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[SH_PROM15:%.*]] = trunc <8 x i32> [[SPLAT_SPLAT14]] to <8 x i16>
-// CHECK:   [[SHL16:%.*]] = shl <8 x i16> [[TMP19]], [[SH_PROM15]]
-// CHECK:   store volatile <8 x i16> [[SHL16]], ptr @ss, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[SHL17:%.*]] = shl <8 x i16> [[TMP20]], splat (i16 5)
-// CHECK:   store volatile <8 x i16> [[SHL17]], ptr @ss, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[SHL18:%.*]] = shl <8 x i16> [[TMP22]], [[TMP21]]
-// CHECK:   store volatile <8 x i16> [[SHL18]], ptr @us, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[SHL19:%.*]] = shl <8 x i16> [[TMP24]], [[TMP23]]
-// CHECK:   store volatile <8 x i16> [[SHL19]], ptr @us, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT20:%.*]] = insertelement <8 x i32> poison, i32 [[TMP25]], i64 0
-// CHECK:   [[SPLAT_SPLAT21:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT20]], <8 x i32> poison, <8 x i32> zeroinitializer
-// CHECK:   [[TMP26:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[SH_PROM22:%.*]] = trunc <8 x i32> [[SPLAT_SPLAT21]] to <8 x i16>
-// CHECK:   [[SHL23:%.*]] = shl <8 x i16> [[TMP26]], [[SH_PROM22]]
-// CHECK:   store volatile <8 x i16> [[SHL23]], ptr @us, align 8
-// CHECK:   [[TMP27:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[SHL24:%.*]] = shl <8 x i16> [[TMP27]], splat (i16 5)
-// CHECK:   store volatile <8 x i16> [[SHL24]], ptr @us, align 8
-// CHECK:   [[TMP28:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[SHL25:%.*]] = shl <4 x i32> [[TMP29]], [[TMP28]]
-// CHECK:   store volatile <4 x i32> [[SHL25]], ptr @si, align 8
-// CHECK:   [[TMP30:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[TMP31:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[SHL26:%.*]] = shl <4 x i32> [[TMP31]], [[TMP30]]
-// CHECK:   store volatile <4 x i32> [[SHL26]], ptr @si, align 8
-// CHECK:   [[TMP32:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT27:%.*]] = insertelement <4 x i32> poison, i32 [[TMP32]], i64 0
-// CHECK:   [[SPLAT_SPLAT28:%.*]] = shufflevector <4 x i32> [[SPLAT_SPLATINSERT27]], <4 x i32> poison, <4 x i32> zeroinitializer
-// CHECK:   [[TMP33:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[SHL29:%.*]] = shl <4 x i32> [[TMP33]], [[SPLAT_SPLAT28]]
-// CHECK:   store volatile <4 x i32> [[SHL29]], ptr @si, align 8
-// CHECK:   [[TMP34:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[SHL30:%.*]] = shl <4 x i32> [[TMP34]], splat (i32 5)
-// CHECK:   store volatile <4 x i32> [[SHL30]], ptr @si, align 8
-// CHECK:   [[TMP35:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[TMP36:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[SHL31:%.*]] = shl <4 x i32> [[TMP36]], [[TMP35]]
-// CHECK:   store volatile <4 x i32> [[SHL31]], ptr @ui, align 8
-// CHECK:   [[TMP37:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[TMP38:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[SHL32:%.*]] = shl <4 x i32> [[TMP38]], [[TMP37]]
-// CHECK:   store volatile <4 x i32> [[SHL32]], ptr @ui, align 8
-// CHECK:   [[TMP39:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT33:%.*]] = insertelement <4 x i32> poison, i32 [[TMP39]], i64 0
-// CHECK:   [[SPLAT_SPLAT34:%.*]] = shufflevector <4 x i32> [[SPLAT_SPLATINSERT33]], <4 x i32> poison, <4 x i32> zeroinitializer
-// CHECK:   [[TMP40:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[SHL35:%.*]] = shl <4 x i32> [[TMP40]], [[SPLAT_SPLAT34]]
-// CHECK:   store volatile <4 x i32> [[SHL35]], ptr @ui, align 8
-// CHECK:   [[TMP41:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[SHL36:%.*]] = shl <4 x i32> [[TMP41]], splat (i32 5)
-// CHECK:   store volatile <4 x i32> [[SHL36]], ptr @ui, align 8
-// CHECK:   [[TMP42:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[TMP43:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[SHL37:%.*]] = shl <2 x i64> [[TMP43]], [[TMP42]]
-// CHECK:   store volatile <2 x i64> [[SHL37]], ptr @sl, align 8
-// CHECK:   [[TMP44:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[TMP45:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[SHL38:%.*]] = shl <2 x i64> [[TMP45]], [[TMP44]]
-// CHECK:   store volatile <2 x i64> [[SHL38]], ptr @sl, align 8
-// CHECK:   [[TMP46:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT39:%.*]] = insertelement <2 x i32> poison, i32 [[TMP46]], i64 0
-// CHECK:   [[SPLAT_SPLAT40:%.*]] = shufflevector <2 x i32> [[SPLAT_SPLATINSERT39]], <2 x i32> poison, <2 x i32> zeroinitializer
-// CHECK:   [[TMP47:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[SH_PROM41:%.*]] = zext <2 x i32> [[SPLAT_SPLAT40]] to <2 x i64>
-// CHECK:   [[SHL42:%.*]] = shl <2 x i64> [[TMP47]], [[SH_PROM41]]
-// CHECK:   store volatile <2 x i64> [[SHL42]], ptr @sl, align 8
-// CHECK:   [[TMP48:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[SHL43:%.*]] = shl <2 x i64> [[TMP48]], splat (i64 5)
-// CHECK:   store volatile <2 x i64> [[SHL43]], ptr @sl, align 8
-// CHECK:   [[TMP49:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[TMP50:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[SHL44:%.*]] = shl <2 x i64> [[TMP50]], [[TMP49]]
-// CHECK:   store volatile <2 x i64> [[SHL44]], ptr @ul, align 8
-// CHECK:   [[TMP51:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[TMP52:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[SHL45:%.*]] = shl <2 x i64> [[TMP52]], [[TMP51]]
-// CHECK:   store volatile <2 x i64> [[SHL45]], ptr @ul, align 8
-// CHECK:   [[TMP53:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT46:%.*]] = insertelement <2 x i32> poison, i32 [[TMP53]], i64 0
-// CHECK:   [[SPLAT_SPLAT47:%.*]] = shufflevector <2 x i32> [[SPLAT_SPLATINSERT46]], <2 x i32> poison, <2 x i32> zeroinitializer
-// CHECK:   [[TMP54:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[SH_PROM48:%.*]] = zext <2 x i32> [[SPLAT_SPLAT47]] to <2 x i64>
-// CHECK:   [[SHL49:%.*]] = shl <2 x i64> [[TMP54]], [[SH_PROM48]]
-// CHECK:   store volatile <2 x i64> [[SHL49]], ptr @ul, align 8
-// CHECK:   [[TMP55:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[SHL50:%.*]] = shl <2 x i64> [[TMP55]], splat (i64 5)
-// CHECK:   store volatile <2 x i64> [[SHL50]], ptr @ul, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_sl_assign(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[SHL:%.*]] = shl <16 x i8> [[TMP1]], [[TMP0]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHL]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[SHL1:%.*]] = shl <16 x i8> [[TMP3]], [[TMP2]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHL1]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <16 x i32> poison, i32 [[TMP4]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <16 x i32> [[SPLAT_SPLATINSERT]], <16 x i32> poison, <16 x i32> zeroinitializer
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[SH_PROM:%.*]] = trunc <16 x i32> [[SPLAT_SPLAT]] to <16 x i8>
+// CHECK-NEXT:    [[SHL2:%.*]] = shl <16 x i8> [[TMP5]], [[SH_PROM]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHL2]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[SHL3:%.*]] = shl <16 x i8> [[TMP6]], splat (i8 5)
+// CHECK-NEXT:    store volatile <16 x i8> [[SHL3]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[SHL4:%.*]] = shl <16 x i8> [[TMP8]], [[TMP7]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHL4]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[SHL5:%.*]] = shl <16 x i8> [[TMP10]], [[TMP9]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHL5]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT6:%.*]] = insertelement <16 x i32> poison, i32 [[TMP11]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT7:%.*]] = shufflevector <16 x i32> [[SPLAT_SPLATINSERT6]], <16 x i32> poison, <16 x i32> zeroinitializer
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[SH_PROM8:%.*]] = trunc <16 x i32> [[SPLAT_SPLAT7]] to <16 x i8>
+// CHECK-NEXT:    [[SHL9:%.*]] = shl <16 x i8> [[TMP12]], [[SH_PROM8]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHL9]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[SHL10:%.*]] = shl <16 x i8> [[TMP13]], splat (i8 5)
+// CHECK-NEXT:    store volatile <16 x i8> [[SHL10]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[SHL11:%.*]] = shl <8 x i16> [[TMP15]], [[TMP14]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHL11]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[SHL12:%.*]] = shl <8 x i16> [[TMP17]], [[TMP16]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHL12]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT13:%.*]] = insertelement <8 x i32> poison, i32 [[TMP18]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT14:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT13]], <8 x i32> poison, <8 x i32> zeroinitializer
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[SH_PROM15:%.*]] = trunc <8 x i32> [[SPLAT_SPLAT14]] to <8 x i16>
+// CHECK-NEXT:    [[SHL16:%.*]] = shl <8 x i16> [[TMP19]], [[SH_PROM15]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHL16]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[SHL17:%.*]] = shl <8 x i16> [[TMP20]], splat (i16 5)
+// CHECK-NEXT:    store volatile <8 x i16> [[SHL17]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[SHL18:%.*]] = shl <8 x i16> [[TMP22]], [[TMP21]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHL18]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[SHL19:%.*]] = shl <8 x i16> [[TMP24]], [[TMP23]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHL19]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT20:%.*]] = insertelement <8 x i32> poison, i32 [[TMP25]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT21:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT20]], <8 x i32> poison, <8 x i32> zeroinitializer
+// CHECK-NEXT:    [[TMP26:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[SH_PROM22:%.*]] = trunc <8 x i32> [[SPLAT_SPLAT21]] to <8 x i16>
+// CHECK-NEXT:    [[SHL23:%.*]] = shl <8 x i16> [[TMP26]], [[SH_PROM22]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHL23]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP27:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[SHL24:%.*]] = shl <8 x i16> [[TMP27]], splat (i16 5)
+// CHECK-NEXT:    store volatile <8 x i16> [[SHL24]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP28:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[SHL25:%.*]] = shl <4 x i32> [[TMP29]], [[TMP28]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHL25]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP30:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[TMP31:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[SHL26:%.*]] = shl <4 x i32> [[TMP31]], [[TMP30]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHL26]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP32:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT27:%.*]] = insertelement <4 x i32> poison, i32 [[TMP32]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT28:%.*]] = shufflevector <4 x i32> [[SPLAT_SPLATINSERT27]], <4 x i32> poison, <4 x i32> zeroinitializer
+// CHECK-NEXT:    [[TMP33:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[SHL29:%.*]] = shl <4 x i32> [[TMP33]], [[SPLAT_SPLAT28]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHL29]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP34:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[SHL30:%.*]] = shl <4 x i32> [[TMP34]], splat (i32 5)
+// CHECK-NEXT:    store volatile <4 x i32> [[SHL30]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP35:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[TMP36:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[SHL31:%.*]] = shl <4 x i32> [[TMP36]], [[TMP35]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHL31]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP37:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[TMP38:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[SHL32:%.*]] = shl <4 x i32> [[TMP38]], [[TMP37]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHL32]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP39:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT33:%.*]] = insertelement <4 x i32> poison, i32 [[TMP39]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT34:%.*]] = shufflevector <4 x i32> [[SPLAT_SPLATINSERT33]], <4 x i32> poison, <4 x i32> zeroinitializer
+// CHECK-NEXT:    [[TMP40:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[SHL35:%.*]] = shl <4 x i32> [[TMP40]], [[SPLAT_SPLAT34]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHL35]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP41:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[SHL36:%.*]] = shl <4 x i32> [[TMP41]], splat (i32 5)
+// CHECK-NEXT:    store volatile <4 x i32> [[SHL36]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP42:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[TMP43:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[SHL37:%.*]] = shl <2 x i64> [[TMP43]], [[TMP42]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHL37]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP44:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[TMP45:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[SHL38:%.*]] = shl <2 x i64> [[TMP45]], [[TMP44]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHL38]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP46:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT39:%.*]] = insertelement <2 x i32> poison, i32 [[TMP46]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT40:%.*]] = shufflevector <2 x i32> [[SPLAT_SPLATINSERT39]], <2 x i32> poison, <2 x i32> zeroinitializer
+// CHECK-NEXT:    [[TMP47:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[SH_PROM41:%.*]] = zext <2 x i32> [[SPLAT_SPLAT40]] to <2 x i64>
+// CHECK-NEXT:    [[SHL42:%.*]] = shl <2 x i64> [[TMP47]], [[SH_PROM41]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHL42]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP48:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[SHL43:%.*]] = shl <2 x i64> [[TMP48]], splat (i64 5)
+// CHECK-NEXT:    store volatile <2 x i64> [[SHL43]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP49:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[TMP50:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[SHL44:%.*]] = shl <2 x i64> [[TMP50]], [[TMP49]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHL44]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP51:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[TMP52:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[SHL45:%.*]] = shl <2 x i64> [[TMP52]], [[TMP51]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHL45]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP53:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT46:%.*]] = insertelement <2 x i32> poison, i32 [[TMP53]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT47:%.*]] = shufflevector <2 x i32> [[SPLAT_SPLATINSERT46]], <2 x i32> poison, <2 x i32> zeroinitializer
+// CHECK-NEXT:    [[TMP54:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[SH_PROM48:%.*]] = zext <2 x i32> [[SPLAT_SPLAT47]] to <2 x i64>
+// CHECK-NEXT:    [[SHL49:%.*]] = shl <2 x i64> [[TMP54]], [[SH_PROM48]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHL49]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP55:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[SHL50:%.*]] = shl <2 x i64> [[TMP55]], splat (i64 5)
+// CHECK-NEXT:    store volatile <2 x i64> [[SHL50]], ptr @ul, align 8
+// CHECK-NEXT:    ret void
+//
 void test_sl_assign(void) {
 
   sc <<= sc2;
@@ -2281,150 +2360,153 @@ void test_sl_assign(void) {
   ul <<= 5;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_sr() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[SHR:%.*]] = ashr <16 x i8> [[TMP0]], [[TMP1]]
-// CHECK:   store volatile <16 x i8> [[SHR]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[SHR1:%.*]] = ashr <16 x i8> [[TMP2]], [[TMP3]]
-// CHECK:   store volatile <16 x i8> [[SHR1]], ptr @sc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT:%.*]] = insertelement <16 x i32> poison, i32 [[TMP5]], i64 0
-// CHECK:   [[SPLAT_SPLAT:%.*]] = shufflevector <16 x i32> [[SPLAT_SPLAT:%.*]]insert, <16 x i32> poison, <16 x i32> zeroinitializer
-// CHECK:   [[SH_PROM:%.*]] = trunc <16 x i32> [[SPLAT_SPLAT]] to <16 x i8>
-// CHECK:   [[SHR2:%.*]] = ashr <16 x i8> [[TMP4]], [[SH_PROM]]
-// CHECK:   store volatile <16 x i8> [[SHR2]], ptr @sc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[SHR3:%.*]] = ashr <16 x i8> [[TMP6]], splat (i8 5)
-// CHECK:   store volatile <16 x i8> [[SHR3]], ptr @sc, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[SHR4:%.*]] = lshr <16 x i8> [[TMP7]], [[TMP8]]
-// CHECK:   store volatile <16 x i8> [[SHR4]], ptr @uc, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[SHR5:%.*]] = lshr <16 x i8> [[TMP9]], [[TMP10]]
-// CHECK:   store volatile <16 x i8> [[SHR5]], ptr @uc, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT6:%.*]] = insertelement <16 x i32> poison, i32 [[TMP12]], i64 0
-// CHECK:   [[SPLAT_SPLAT7:%.*]] = shufflevector <16 x i32> [[SPLAT_SPLATINSERT6]], <16 x i32> poison, <16 x i32> zeroinitializer
-// CHECK:   [[SH_PROM8:%.*]] = trunc <16 x i32> [[SPLAT_SPLAT7]] to <16 x i8>
-// CHECK:   [[SHR9:%.*]] = lshr <16 x i8> [[TMP11]], [[SH_PROM8]]
-// CHECK:   store volatile <16 x i8> [[SHR9]], ptr @uc, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[SHR10:%.*]] = lshr <16 x i8> [[TMP13]], splat (i8 5)
-// CHECK:   store volatile <16 x i8> [[SHR10]], ptr @uc, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[SHR11:%.*]] = ashr <8 x i16> [[TMP14]], [[TMP15]]
-// CHECK:   store volatile <8 x i16> [[SHR11]], ptr @ss, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[SHR12:%.*]] = ashr <8 x i16> [[TMP16]], [[TMP17]]
-// CHECK:   store volatile <8 x i16> [[SHR12]], ptr @ss, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP19:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT13:%.*]] = insertelement <8 x i32> poison, i32 [[TMP19]], i64 0
-// CHECK:   [[SPLAT_SPLAT14:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT13]], <8 x i32> poison, <8 x i32> zeroinitializer
-// CHECK:   [[SH_PROM15:%.*]] = trunc <8 x i32> [[SPLAT_SPLAT14]] to <8 x i16>
-// CHECK:   [[SHR16:%.*]] = ashr <8 x i16> [[TMP18]], [[SH_PROM15]]
-// CHECK:   store volatile <8 x i16> [[SHR16]], ptr @ss, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[SHR17:%.*]] = ashr <8 x i16> [[TMP20]], splat (i16 5)
-// CHECK:   store volatile <8 x i16> [[SHR17]], ptr @ss, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[SHR18:%.*]] = lshr <8 x i16> [[TMP21]], [[TMP22]]
-// CHECK:   store volatile <8 x i16> [[SHR18]], ptr @us, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[SHR19:%.*]] = lshr <8 x i16> [[TMP23]], [[TMP24]]
-// CHECK:   store volatile <8 x i16> [[SHR19]], ptr @us, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP26:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT20:%.*]] = insertelement <8 x i32> poison, i32 [[TMP26]], i64 0
-// CHECK:   [[SPLAT_SPLAT21:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT20]], <8 x i32> poison, <8 x i32> zeroinitializer
-// CHECK:   [[SH_PROM22:%.*]] = trunc <8 x i32> [[SPLAT_SPLAT21]] to <8 x i16>
-// CHECK:   [[SHR23:%.*]] = lshr <8 x i16> [[TMP25]], [[SH_PROM22]]
-// CHECK:   store volatile <8 x i16> [[SHR23]], ptr @us, align 8
-// CHECK:   [[TMP27:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[SHR24:%.*]] = lshr <8 x i16> [[TMP27]], splat (i16 5)
-// CHECK:   store volatile <8 x i16> [[SHR24]], ptr @us, align 8
-// CHECK:   [[TMP28:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[SHR25:%.*]] = ashr <4 x i32> [[TMP28]], [[TMP29]]
-// CHECK:   store volatile <4 x i32> [[SHR25]], ptr @si, align 8
-// CHECK:   [[TMP30:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP31:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[SHR26:%.*]] = ashr <4 x i32> [[TMP30]], [[TMP31]]
-// CHECK:   store volatile <4 x i32> [[SHR26]], ptr @si, align 8
-// CHECK:   [[TMP32:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP33:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT27:%.*]] = insertelement <4 x i32> poison, i32 [[TMP33]], i64 0
-// CHECK:   [[SPLAT_SPLAT28:%.*]] = shufflevector <4 x i32> [[SPLAT_SPLATINSERT27]], <4 x i32> poison, <4 x i32> zeroinitializer
-// CHECK:   [[SHR29:%.*]] = ashr <4 x i32> [[TMP32]], [[SPLAT_SPLAT28]]
-// CHECK:   store volatile <4 x i32> [[SHR29]], ptr @si, align 8
-// CHECK:   [[TMP34:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[SHR30:%.*]] = ashr <4 x i32> [[TMP34]], splat (i32 5)
-// CHECK:   store volatile <4 x i32> [[SHR30]], ptr @si, align 8
-// CHECK:   [[TMP35:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP36:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[SHR31:%.*]] = lshr <4 x i32> [[TMP35]], [[TMP36]]
-// CHECK:   store volatile <4 x i32> [[SHR31]], ptr @ui, align 8
-// CHECK:   [[TMP37:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP38:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[SHR32:%.*]] = lshr <4 x i32> [[TMP37]], [[TMP38]]
-// CHECK:   store volatile <4 x i32> [[SHR32]], ptr @ui, align 8
-// CHECK:   [[TMP39:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP40:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT33:%.*]] = insertelement <4 x i32> poison, i32 [[TMP40]], i64 0
-// CHECK:   [[SPLAT_SPLAT34:%.*]] = shufflevector <4 x i32> [[SPLAT_SPLATINSERT33]], <4 x i32> poison, <4 x i32> zeroinitializer
-// CHECK:   [[SHR35:%.*]] = lshr <4 x i32> [[TMP39]], [[SPLAT_SPLAT34]]
-// CHECK:   store volatile <4 x i32> [[SHR35]], ptr @ui, align 8
-// CHECK:   [[TMP41:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[SHR36:%.*]] = lshr <4 x i32> [[TMP41]], splat (i32 5)
-// CHECK:   store volatile <4 x i32> [[SHR36]], ptr @ui, align 8
-// CHECK:   [[TMP42:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP43:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[SHR37:%.*]] = ashr <2 x i64> [[TMP42]], [[TMP43]]
-// CHECK:   store volatile <2 x i64> [[SHR37]], ptr @sl, align 8
-// CHECK:   [[TMP44:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP45:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[SHR38:%.*]] = ashr <2 x i64> [[TMP44]], [[TMP45]]
-// CHECK:   store volatile <2 x i64> [[SHR38]], ptr @sl, align 8
-// CHECK:   [[TMP46:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP47:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT39:%.*]] = insertelement <2 x i32> poison, i32 [[TMP47]], i64 0
-// CHECK:   [[SPLAT_SPLAT40:%.*]] = shufflevector <2 x i32> [[SPLAT_SPLATINSERT39]], <2 x i32> poison, <2 x i32> zeroinitializer
-// CHECK:   [[SH_PROM41:%.*]] = zext <2 x i32> [[SPLAT_SPLAT40]] to <2 x i64>
-// CHECK:   [[SHR42:%.*]] = ashr <2 x i64> [[TMP46]], [[SH_PROM41]]
-// CHECK:   store volatile <2 x i64> [[SHR42]], ptr @sl, align 8
-// CHECK:   [[TMP48:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[SHR43:%.*]] = ashr <2 x i64> [[TMP48]], splat (i64 5)
-// CHECK:   store volatile <2 x i64> [[SHR43]], ptr @sl, align 8
-// CHECK:   [[TMP49:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP50:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[SHR44:%.*]] = lshr <2 x i64> [[TMP49]], [[TMP50]]
-// CHECK:   store volatile <2 x i64> [[SHR44]], ptr @ul, align 8
-// CHECK:   [[TMP51:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP52:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[SHR45:%.*]] = lshr <2 x i64> [[TMP51]], [[TMP52]]
-// CHECK:   store volatile <2 x i64> [[SHR45]], ptr @ul, align 8
-// CHECK:   [[TMP53:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP54:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT46:%.*]] = insertelement <2 x i32> poison, i32 [[TMP54]], i64 0
-// CHECK:   [[SPLAT_SPLAT47:%.*]] = shufflevector <2 x i32> [[SPLAT_SPLATINSERT46]], <2 x i32> poison, <2 x i32> zeroinitializer
-// CHECK:   [[SH_PROM48:%.*]] = zext <2 x i32> [[SPLAT_SPLAT47]] to <2 x i64>
-// CHECK:   [[SHR49:%.*]] = lshr <2 x i64> [[TMP53]], [[SH_PROM48]]
-// CHECK:   store volatile <2 x i64> [[SHR49]], ptr @ul, align 8
-// CHECK:   [[TMP55:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[SHR50:%.*]] = lshr <2 x i64> [[TMP55]], splat (i64 5)
-// CHECK:   store volatile <2 x i64> [[SHR50]], ptr @ul, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_sr(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[SHR:%.*]] = ashr <16 x i8> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHR]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[SHR1:%.*]] = ashr <16 x i8> [[TMP2]], [[TMP3]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHR1]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <16 x i32> poison, i32 [[TMP5]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <16 x i32> [[SPLAT_SPLATINSERT]], <16 x i32> poison, <16 x i32> zeroinitializer
+// CHECK-NEXT:    [[SH_PROM:%.*]] = trunc <16 x i32> [[SPLAT_SPLAT]] to <16 x i8>
+// CHECK-NEXT:    [[SHR2:%.*]] = ashr <16 x i8> [[TMP4]], [[SH_PROM]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHR2]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[SHR3:%.*]] = ashr <16 x i8> [[TMP6]], splat (i8 5)
+// CHECK-NEXT:    store volatile <16 x i8> [[SHR3]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[SHR4:%.*]] = lshr <16 x i8> [[TMP7]], [[TMP8]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHR4]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[SHR5:%.*]] = lshr <16 x i8> [[TMP9]], [[TMP10]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHR5]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT6:%.*]] = insertelement <16 x i32> poison, i32 [[TMP12]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT7:%.*]] = shufflevector <16 x i32> [[SPLAT_SPLATINSERT6]], <16 x i32> poison, <16 x i32> zeroinitializer
+// CHECK-NEXT:    [[SH_PROM8:%.*]] = trunc <16 x i32> [[SPLAT_SPLAT7]] to <16 x i8>
+// CHECK-NEXT:    [[SHR9:%.*]] = lshr <16 x i8> [[TMP11]], [[SH_PROM8]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHR9]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[SHR10:%.*]] = lshr <16 x i8> [[TMP13]], splat (i8 5)
+// CHECK-NEXT:    store volatile <16 x i8> [[SHR10]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[SHR11:%.*]] = ashr <8 x i16> [[TMP14]], [[TMP15]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHR11]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[SHR12:%.*]] = ashr <8 x i16> [[TMP16]], [[TMP17]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHR12]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT13:%.*]] = insertelement <8 x i32> poison, i32 [[TMP19]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT14:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT13]], <8 x i32> poison, <8 x i32> zeroinitializer
+// CHECK-NEXT:    [[SH_PROM15:%.*]] = trunc <8 x i32> [[SPLAT_SPLAT14]] to <8 x i16>
+// CHECK-NEXT:    [[SHR16:%.*]] = ashr <8 x i16> [[TMP18]], [[SH_PROM15]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHR16]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[SHR17:%.*]] = ashr <8 x i16> [[TMP20]], splat (i16 5)
+// CHECK-NEXT:    store volatile <8 x i16> [[SHR17]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[SHR18:%.*]] = lshr <8 x i16> [[TMP21]], [[TMP22]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHR18]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[SHR19:%.*]] = lshr <8 x i16> [[TMP23]], [[TMP24]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHR19]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP26:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT20:%.*]] = insertelement <8 x i32> poison, i32 [[TMP26]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT21:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT20]], <8 x i32> poison, <8 x i32> zeroinitializer
+// CHECK-NEXT:    [[SH_PROM22:%.*]] = trunc <8 x i32> [[SPLAT_SPLAT21]] to <8 x i16>
+// CHECK-NEXT:    [[SHR23:%.*]] = lshr <8 x i16> [[TMP25]], [[SH_PROM22]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHR23]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP27:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[SHR24:%.*]] = lshr <8 x i16> [[TMP27]], splat (i16 5)
+// CHECK-NEXT:    store volatile <8 x i16> [[SHR24]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP28:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[SHR25:%.*]] = ashr <4 x i32> [[TMP28]], [[TMP29]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHR25]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP30:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP31:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[SHR26:%.*]] = ashr <4 x i32> [[TMP30]], [[TMP31]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHR26]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP32:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP33:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT27:%.*]] = insertelement <4 x i32> poison, i32 [[TMP33]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT28:%.*]] = shufflevector <4 x i32> [[SPLAT_SPLATINSERT27]], <4 x i32> poison, <4 x i32> zeroinitializer
+// CHECK-NEXT:    [[SHR29:%.*]] = ashr <4 x i32> [[TMP32]], [[SPLAT_SPLAT28]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHR29]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP34:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[SHR30:%.*]] = ashr <4 x i32> [[TMP34]], splat (i32 5)
+// CHECK-NEXT:    store volatile <4 x i32> [[SHR30]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP35:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP36:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[SHR31:%.*]] = lshr <4 x i32> [[TMP35]], [[TMP36]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHR31]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP37:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP38:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[SHR32:%.*]] = lshr <4 x i32> [[TMP37]], [[TMP38]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHR32]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP39:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP40:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT33:%.*]] = insertelement <4 x i32> poison, i32 [[TMP40]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT34:%.*]] = shufflevector <4 x i32> [[SPLAT_SPLATINSERT33]], <4 x i32> poison, <4 x i32> zeroinitializer
+// CHECK-NEXT:    [[SHR35:%.*]] = lshr <4 x i32> [[TMP39]], [[SPLAT_SPLAT34]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHR35]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP41:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[SHR36:%.*]] = lshr <4 x i32> [[TMP41]], splat (i32 5)
+// CHECK-NEXT:    store volatile <4 x i32> [[SHR36]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP42:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP43:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[SHR37:%.*]] = ashr <2 x i64> [[TMP42]], [[TMP43]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHR37]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP44:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP45:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[SHR38:%.*]] = ashr <2 x i64> [[TMP44]], [[TMP45]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHR38]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP46:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP47:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT39:%.*]] = insertelement <2 x i32> poison, i32 [[TMP47]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT40:%.*]] = shufflevector <2 x i32> [[SPLAT_SPLATINSERT39]], <2 x i32> poison, <2 x i32> zeroinitializer
+// CHECK-NEXT:    [[SH_PROM41:%.*]] = zext <2 x i32> [[SPLAT_SPLAT40]] to <2 x i64>
+// CHECK-NEXT:    [[SHR42:%.*]] = ashr <2 x i64> [[TMP46]], [[SH_PROM41]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHR42]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP48:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[SHR43:%.*]] = ashr <2 x i64> [[TMP48]], splat (i64 5)
+// CHECK-NEXT:    store volatile <2 x i64> [[SHR43]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP49:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP50:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[SHR44:%.*]] = lshr <2 x i64> [[TMP49]], [[TMP50]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHR44]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP51:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP52:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[SHR45:%.*]] = lshr <2 x i64> [[TMP51]], [[TMP52]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHR45]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP53:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP54:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT46:%.*]] = insertelement <2 x i32> poison, i32 [[TMP54]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT47:%.*]] = shufflevector <2 x i32> [[SPLAT_SPLATINSERT46]], <2 x i32> poison, <2 x i32> zeroinitializer
+// CHECK-NEXT:    [[SH_PROM48:%.*]] = zext <2 x i32> [[SPLAT_SPLAT47]] to <2 x i64>
+// CHECK-NEXT:    [[SHR49:%.*]] = lshr <2 x i64> [[TMP53]], [[SH_PROM48]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHR49]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP55:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[SHR50:%.*]] = lshr <2 x i64> [[TMP55]], splat (i64 5)
+// CHECK-NEXT:    store volatile <2 x i64> [[SHR50]], ptr @ul, align 8
+// CHECK-NEXT:    ret void
+//
 void test_sr(void) {
 
   sc = sc >> sc2;
@@ -2464,150 +2546,153 @@ void test_sr(void) {
   ul = ul >> 5;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_sr_assign() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[SHR:%.*]] = ashr <16 x i8> [[TMP1]], [[TMP0]]
-// CHECK:   store volatile <16 x i8> [[SHR]], ptr @sc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[SHR1:%.*]] = ashr <16 x i8> [[TMP3]], [[TMP2]]
-// CHECK:   store volatile <16 x i8> [[SHR1]], ptr @sc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT:%.*]] = insertelement <16 x i32> poison, i32 [[TMP4]], i64 0
-// CHECK:   [[SPLAT_SPLAT:%.*]] = shufflevector <16 x i32> [[SPLAT_SPLAT:%.*]]insert, <16 x i32> poison, <16 x i32> zeroinitializer
-// CHECK:   [[TMP5:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[SH_PROM:%.*]] = trunc <16 x i32> [[SPLAT_SPLAT]] to <16 x i8>
-// CHECK:   [[SHR2:%.*]] = ashr <16 x i8> [[TMP5]], [[SH_PROM]]
-// CHECK:   store volatile <16 x i8> [[SHR2]], ptr @sc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[SHR3:%.*]] = ashr <16 x i8> [[TMP6]], splat (i8 5)
-// CHECK:   store volatile <16 x i8> [[SHR3]], ptr @sc, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[SHR4:%.*]] = lshr <16 x i8> [[TMP8]], [[TMP7]]
-// CHECK:   store volatile <16 x i8> [[SHR4]], ptr @uc, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[SHR5:%.*]] = lshr <16 x i8> [[TMP10]], [[TMP9]]
-// CHECK:   store volatile <16 x i8> [[SHR5]], ptr @uc, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT6:%.*]] = insertelement <16 x i32> poison, i32 [[TMP11]], i64 0
-// CHECK:   [[SPLAT_SPLAT7:%.*]] = shufflevector <16 x i32> [[SPLAT_SPLATINSERT6]], <16 x i32> poison, <16 x i32> zeroinitializer
-// CHECK:   [[TMP12:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[SH_PROM8:%.*]] = trunc <16 x i32> [[SPLAT_SPLAT7]] to <16 x i8>
-// CHECK:   [[SHR9:%.*]] = lshr <16 x i8> [[TMP12]], [[SH_PROM8]]
-// CHECK:   store volatile <16 x i8> [[SHR9]], ptr @uc, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[SHR10:%.*]] = lshr <16 x i8> [[TMP13]], splat (i8 5)
-// CHECK:   store volatile <16 x i8> [[SHR10]], ptr @uc, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[SHR11:%.*]] = ashr <8 x i16> [[TMP15]], [[TMP14]]
-// CHECK:   store volatile <8 x i16> [[SHR11]], ptr @ss, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[SHR12:%.*]] = ashr <8 x i16> [[TMP17]], [[TMP16]]
-// CHECK:   store volatile <8 x i16> [[SHR12]], ptr @ss, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT13:%.*]] = insertelement <8 x i32> poison, i32 [[TMP18]], i64 0
-// CHECK:   [[SPLAT_SPLAT14:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT13]], <8 x i32> poison, <8 x i32> zeroinitializer
-// CHECK:   [[TMP19:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[SH_PROM15:%.*]] = trunc <8 x i32> [[SPLAT_SPLAT14]] to <8 x i16>
-// CHECK:   [[SHR16:%.*]] = ashr <8 x i16> [[TMP19]], [[SH_PROM15]]
-// CHECK:   store volatile <8 x i16> [[SHR16]], ptr @ss, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[SHR17:%.*]] = ashr <8 x i16> [[TMP20]], splat (i16 5)
-// CHECK:   store volatile <8 x i16> [[SHR17]], ptr @ss, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[SHR18:%.*]] = lshr <8 x i16> [[TMP22]], [[TMP21]]
-// CHECK:   store volatile <8 x i16> [[SHR18]], ptr @us, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[SHR19:%.*]] = lshr <8 x i16> [[TMP24]], [[TMP23]]
-// CHECK:   store volatile <8 x i16> [[SHR19]], ptr @us, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT20:%.*]] = insertelement <8 x i32> poison, i32 [[TMP25]], i64 0
-// CHECK:   [[SPLAT_SPLAT21:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT20]], <8 x i32> poison, <8 x i32> zeroinitializer
-// CHECK:   [[TMP26:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[SH_PROM22:%.*]] = trunc <8 x i32> [[SPLAT_SPLAT21]] to <8 x i16>
-// CHECK:   [[SHR23:%.*]] = lshr <8 x i16> [[TMP26]], [[SH_PROM22]]
-// CHECK:   store volatile <8 x i16> [[SHR23]], ptr @us, align 8
-// CHECK:   [[TMP27:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[SHR24:%.*]] = lshr <8 x i16> [[TMP27]], splat (i16 5)
-// CHECK:   store volatile <8 x i16> [[SHR24]], ptr @us, align 8
-// CHECK:   [[TMP28:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[SHR25:%.*]] = ashr <4 x i32> [[TMP29]], [[TMP28]]
-// CHECK:   store volatile <4 x i32> [[SHR25]], ptr @si, align 8
-// CHECK:   [[TMP30:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[TMP31:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[SHR26:%.*]] = ashr <4 x i32> [[TMP31]], [[TMP30]]
-// CHECK:   store volatile <4 x i32> [[SHR26]], ptr @si, align 8
-// CHECK:   [[TMP32:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT27:%.*]] = insertelement <4 x i32> poison, i32 [[TMP32]], i64 0
-// CHECK:   [[SPLAT_SPLAT28:%.*]] = shufflevector <4 x i32> [[SPLAT_SPLATINSERT27]], <4 x i32> poison, <4 x i32> zeroinitializer
-// CHECK:   [[TMP33:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[SHR29:%.*]] = ashr <4 x i32> [[TMP33]], [[SPLAT_SPLAT28]]
-// CHECK:   store volatile <4 x i32> [[SHR29]], ptr @si, align 8
-// CHECK:   [[TMP34:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[SHR30:%.*]] = ashr <4 x i32> [[TMP34]], splat (i32 5)
-// CHECK:   store volatile <4 x i32> [[SHR30]], ptr @si, align 8
-// CHECK:   [[TMP35:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[TMP36:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[SHR31:%.*]] = lshr <4 x i32> [[TMP36]], [[TMP35]]
-// CHECK:   store volatile <4 x i32> [[SHR31]], ptr @ui, align 8
-// CHECK:   [[TMP37:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[TMP38:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[SHR32:%.*]] = lshr <4 x i32> [[TMP38]], [[TMP37]]
-// CHECK:   store volatile <4 x i32> [[SHR32]], ptr @ui, align 8
-// CHECK:   [[TMP39:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT33:%.*]] = insertelement <4 x i32> poison, i32 [[TMP39]], i64 0
-// CHECK:   [[SPLAT_SPLAT34:%.*]] = shufflevector <4 x i32> [[SPLAT_SPLATINSERT33]], <4 x i32> poison, <4 x i32> zeroinitializer
-// CHECK:   [[TMP40:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[SHR35:%.*]] = lshr <4 x i32> [[TMP40]], [[SPLAT_SPLAT34]]
-// CHECK:   store volatile <4 x i32> [[SHR35]], ptr @ui, align 8
-// CHECK:   [[TMP41:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[SHR36:%.*]] = lshr <4 x i32> [[TMP41]], splat (i32 5)
-// CHECK:   store volatile <4 x i32> [[SHR36]], ptr @ui, align 8
-// CHECK:   [[TMP42:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[TMP43:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[SHR37:%.*]] = ashr <2 x i64> [[TMP43]], [[TMP42]]
-// CHECK:   store volatile <2 x i64> [[SHR37]], ptr @sl, align 8
-// CHECK:   [[TMP44:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[TMP45:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[SHR38:%.*]] = ashr <2 x i64> [[TMP45]], [[TMP44]]
-// CHECK:   store volatile <2 x i64> [[SHR38]], ptr @sl, align 8
-// CHECK:   [[TMP46:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT39:%.*]] = insertelement <2 x i32> poison, i32 [[TMP46]], i64 0
-// CHECK:   [[SPLAT_SPLAT40:%.*]] = shufflevector <2 x i32> [[SPLAT_SPLATINSERT39]], <2 x i32> poison, <2 x i32> zeroinitializer
-// CHECK:   [[TMP47:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[SH_PROM41:%.*]] = zext <2 x i32> [[SPLAT_SPLAT40]] to <2 x i64>
-// CHECK:   [[SHR42:%.*]] = ashr <2 x i64> [[TMP47]], [[SH_PROM41]]
-// CHECK:   store volatile <2 x i64> [[SHR42]], ptr @sl, align 8
-// CHECK:   [[TMP48:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[SHR43:%.*]] = ashr <2 x i64> [[TMP48]], splat (i64 5)
-// CHECK:   store volatile <2 x i64> [[SHR43]], ptr @sl, align 8
-// CHECK:   [[TMP49:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[TMP50:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[SHR44:%.*]] = lshr <2 x i64> [[TMP50]], [[TMP49]]
-// CHECK:   store volatile <2 x i64> [[SHR44]], ptr @ul, align 8
-// CHECK:   [[TMP51:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[TMP52:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[SHR45:%.*]] = lshr <2 x i64> [[TMP52]], [[TMP51]]
-// CHECK:   store volatile <2 x i64> [[SHR45]], ptr @ul, align 8
-// CHECK:   [[TMP53:%.*]] = load volatile i32, ptr @cnt, align 4
-// CHECK:   [[SPLAT_SPLATINSERT46:%.*]] = insertelement <2 x i32> poison, i32 [[TMP53]], i64 0
-// CHECK:   [[SPLAT_SPLAT47:%.*]] = shufflevector <2 x i32> [[SPLAT_SPLATINSERT46]], <2 x i32> poison, <2 x i32> zeroinitializer
-// CHECK:   [[TMP54:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[SH_PROM48:%.*]] = zext <2 x i32> [[SPLAT_SPLAT47]] to <2 x i64>
-// CHECK:   [[SHR49:%.*]] = lshr <2 x i64> [[TMP54]], [[SH_PROM48]]
-// CHECK:   store volatile <2 x i64> [[SHR49]], ptr @ul, align 8
-// CHECK:   [[TMP55:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[SHR50:%.*]] = lshr <2 x i64> [[TMP55]], splat (i64 5)
-// CHECK:   store volatile <2 x i64> [[SHR50]], ptr @ul, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_sr_assign(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[SHR:%.*]] = ashr <16 x i8> [[TMP1]], [[TMP0]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHR]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[SHR1:%.*]] = ashr <16 x i8> [[TMP3]], [[TMP2]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHR1]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <16 x i32> poison, i32 [[TMP4]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <16 x i32> [[SPLAT_SPLATINSERT]], <16 x i32> poison, <16 x i32> zeroinitializer
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[SH_PROM:%.*]] = trunc <16 x i32> [[SPLAT_SPLAT]] to <16 x i8>
+// CHECK-NEXT:    [[SHR2:%.*]] = ashr <16 x i8> [[TMP5]], [[SH_PROM]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHR2]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[SHR3:%.*]] = ashr <16 x i8> [[TMP6]], splat (i8 5)
+// CHECK-NEXT:    store volatile <16 x i8> [[SHR3]], ptr @sc, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[SHR4:%.*]] = lshr <16 x i8> [[TMP8]], [[TMP7]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHR4]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[SHR5:%.*]] = lshr <16 x i8> [[TMP10]], [[TMP9]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHR5]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT6:%.*]] = insertelement <16 x i32> poison, i32 [[TMP11]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT7:%.*]] = shufflevector <16 x i32> [[SPLAT_SPLATINSERT6]], <16 x i32> poison, <16 x i32> zeroinitializer
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[SH_PROM8:%.*]] = trunc <16 x i32> [[SPLAT_SPLAT7]] to <16 x i8>
+// CHECK-NEXT:    [[SHR9:%.*]] = lshr <16 x i8> [[TMP12]], [[SH_PROM8]]
+// CHECK-NEXT:    store volatile <16 x i8> [[SHR9]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[SHR10:%.*]] = lshr <16 x i8> [[TMP13]], splat (i8 5)
+// CHECK-NEXT:    store volatile <16 x i8> [[SHR10]], ptr @uc, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[SHR11:%.*]] = ashr <8 x i16> [[TMP15]], [[TMP14]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHR11]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[SHR12:%.*]] = ashr <8 x i16> [[TMP17]], [[TMP16]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHR12]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT13:%.*]] = insertelement <8 x i32> poison, i32 [[TMP18]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT14:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT13]], <8 x i32> poison, <8 x i32> zeroinitializer
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[SH_PROM15:%.*]] = trunc <8 x i32> [[SPLAT_SPLAT14]] to <8 x i16>
+// CHECK-NEXT:    [[SHR16:%.*]] = ashr <8 x i16> [[TMP19]], [[SH_PROM15]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHR16]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[SHR17:%.*]] = ashr <8 x i16> [[TMP20]], splat (i16 5)
+// CHECK-NEXT:    store volatile <8 x i16> [[SHR17]], ptr @ss, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[SHR18:%.*]] = lshr <8 x i16> [[TMP22]], [[TMP21]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHR18]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[SHR19:%.*]] = lshr <8 x i16> [[TMP24]], [[TMP23]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHR19]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT20:%.*]] = insertelement <8 x i32> poison, i32 [[TMP25]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT21:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT20]], <8 x i32> poison, <8 x i32> zeroinitializer
+// CHECK-NEXT:    [[TMP26:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[SH_PROM22:%.*]] = trunc <8 x i32> [[SPLAT_SPLAT21]] to <8 x i16>
+// CHECK-NEXT:    [[SHR23:%.*]] = lshr <8 x i16> [[TMP26]], [[SH_PROM22]]
+// CHECK-NEXT:    store volatile <8 x i16> [[SHR23]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP27:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[SHR24:%.*]] = lshr <8 x i16> [[TMP27]], splat (i16 5)
+// CHECK-NEXT:    store volatile <8 x i16> [[SHR24]], ptr @us, align 8
+// CHECK-NEXT:    [[TMP28:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[SHR25:%.*]] = ashr <4 x i32> [[TMP29]], [[TMP28]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHR25]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP30:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[TMP31:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[SHR26:%.*]] = ashr <4 x i32> [[TMP31]], [[TMP30]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHR26]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP32:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT27:%.*]] = insertelement <4 x i32> poison, i32 [[TMP32]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT28:%.*]] = shufflevector <4 x i32> [[SPLAT_SPLATINSERT27]], <4 x i32> poison, <4 x i32> zeroinitializer
+// CHECK-NEXT:    [[TMP33:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[SHR29:%.*]] = ashr <4 x i32> [[TMP33]], [[SPLAT_SPLAT28]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHR29]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP34:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[SHR30:%.*]] = ashr <4 x i32> [[TMP34]], splat (i32 5)
+// CHECK-NEXT:    store volatile <4 x i32> [[SHR30]], ptr @si, align 8
+// CHECK-NEXT:    [[TMP35:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[TMP36:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[SHR31:%.*]] = lshr <4 x i32> [[TMP36]], [[TMP35]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHR31]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP37:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[TMP38:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[SHR32:%.*]] = lshr <4 x i32> [[TMP38]], [[TMP37]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHR32]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP39:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT33:%.*]] = insertelement <4 x i32> poison, i32 [[TMP39]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT34:%.*]] = shufflevector <4 x i32> [[SPLAT_SPLATINSERT33]], <4 x i32> poison, <4 x i32> zeroinitializer
+// CHECK-NEXT:    [[TMP40:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[SHR35:%.*]] = lshr <4 x i32> [[TMP40]], [[SPLAT_SPLAT34]]
+// CHECK-NEXT:    store volatile <4 x i32> [[SHR35]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP41:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[SHR36:%.*]] = lshr <4 x i32> [[TMP41]], splat (i32 5)
+// CHECK-NEXT:    store volatile <4 x i32> [[SHR36]], ptr @ui, align 8
+// CHECK-NEXT:    [[TMP42:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[TMP43:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[SHR37:%.*]] = ashr <2 x i64> [[TMP43]], [[TMP42]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHR37]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP44:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[TMP45:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[SHR38:%.*]] = ashr <2 x i64> [[TMP45]], [[TMP44]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHR38]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP46:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT39:%.*]] = insertelement <2 x i32> poison, i32 [[TMP46]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT40:%.*]] = shufflevector <2 x i32> [[SPLAT_SPLATINSERT39]], <2 x i32> poison, <2 x i32> zeroinitializer
+// CHECK-NEXT:    [[TMP47:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[SH_PROM41:%.*]] = zext <2 x i32> [[SPLAT_SPLAT40]] to <2 x i64>
+// CHECK-NEXT:    [[SHR42:%.*]] = ashr <2 x i64> [[TMP47]], [[SH_PROM41]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHR42]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP48:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[SHR43:%.*]] = ashr <2 x i64> [[TMP48]], splat (i64 5)
+// CHECK-NEXT:    store volatile <2 x i64> [[SHR43]], ptr @sl, align 8
+// CHECK-NEXT:    [[TMP49:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[TMP50:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[SHR44:%.*]] = lshr <2 x i64> [[TMP50]], [[TMP49]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHR44]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP51:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[TMP52:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[SHR45:%.*]] = lshr <2 x i64> [[TMP52]], [[TMP51]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHR45]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP53:%.*]] = load volatile i32, ptr @cnt, align 4
+// CHECK-NEXT:    [[SPLAT_SPLATINSERT46:%.*]] = insertelement <2 x i32> poison, i32 [[TMP53]], i64 0
+// CHECK-NEXT:    [[SPLAT_SPLAT47:%.*]] = shufflevector <2 x i32> [[SPLAT_SPLATINSERT46]], <2 x i32> poison, <2 x i32> zeroinitializer
+// CHECK-NEXT:    [[TMP54:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[SH_PROM48:%.*]] = zext <2 x i32> [[SPLAT_SPLAT47]] to <2 x i64>
+// CHECK-NEXT:    [[SHR49:%.*]] = lshr <2 x i64> [[TMP54]], [[SH_PROM48]]
+// CHECK-NEXT:    store volatile <2 x i64> [[SHR49]], ptr @ul, align 8
+// CHECK-NEXT:    [[TMP55:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[SHR50:%.*]] = lshr <2 x i64> [[TMP55]], splat (i64 5)
+// CHECK-NEXT:    store volatile <2 x i64> [[SHR50]], ptr @ul, align 8
+// CHECK-NEXT:    ret void
+//
 void test_sr_assign(void) {
 
   sc >>= sc2;
@@ -2648,153 +2733,156 @@ void test_sr_assign(void) {
 }
 
 
-// CHECK-LABEL: define{{.*}} void @test_cmpeq() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[CMP:%.*]] = icmp eq <16 x i8> [[TMP0]], [[TMP1]]
-// CHECK:   [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT]], ptr @bc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[CMP1:%.*]] = icmp eq <16 x i8> [[TMP2]], [[TMP3]]
-// CHECK:   [[SEXT2:%.*]] = sext <16 x i1> [[CMP1]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT2]], ptr @bc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[CMP3:%.*]] = icmp eq <16 x i8> [[TMP4]], [[TMP5]]
-// CHECK:   [[SEXT4:%.*]] = sext <16 x i1> [[CMP3]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT4]], ptr @bc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[CMP5:%.*]] = icmp eq <16 x i8> [[TMP6]], [[TMP7]]
-// CHECK:   [[SEXT6:%.*]] = sext <16 x i1> [[CMP5]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT6]], ptr @bc, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[CMP7:%.*]] = icmp eq <16 x i8> [[TMP8]], [[TMP9]]
-// CHECK:   [[SEXT8:%.*]] = sext <16 x i1> [[CMP7]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT8]], ptr @bc, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[CMP9:%.*]] = icmp eq <16 x i8> [[TMP10]], [[TMP11]]
-// CHECK:   [[SEXT10:%.*]] = sext <16 x i1> [[CMP9]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT10]], ptr @bc, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[CMP11:%.*]] = icmp eq <16 x i8> [[TMP12]], [[TMP13]]
-// CHECK:   [[SEXT12:%.*]] = sext <16 x i1> [[CMP11]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT12]], ptr @bc, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[CMP13:%.*]] = icmp eq <8 x i16> [[TMP14]], [[TMP15]]
-// CHECK:   [[SEXT14:%.*]] = sext <8 x i1> [[CMP13]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT14]], ptr @bs, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[CMP15:%.*]] = icmp eq <8 x i16> [[TMP16]], [[TMP17]]
-// CHECK:   [[SEXT16:%.*]] = sext <8 x i1> [[CMP15]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT16]], ptr @bs, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP19:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[CMP17:%.*]] = icmp eq <8 x i16> [[TMP18]], [[TMP19]]
-// CHECK:   [[SEXT18:%.*]] = sext <8 x i1> [[CMP17]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT18]], ptr @bs, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[CMP19:%.*]] = icmp eq <8 x i16> [[TMP20]], [[TMP21]]
-// CHECK:   [[SEXT20:%.*]] = sext <8 x i1> [[CMP19]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT20]], ptr @bs, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[CMP21:%.*]] = icmp eq <8 x i16> [[TMP22]], [[TMP23]]
-// CHECK:   [[SEXT22:%.*]] = sext <8 x i1> [[CMP21]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT22]], ptr @bs, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[CMP23:%.*]] = icmp eq <8 x i16> [[TMP24]], [[TMP25]]
-// CHECK:   [[SEXT24:%.*]] = sext <8 x i1> [[CMP23]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT24]], ptr @bs, align 8
-// CHECK:   [[TMP26:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP27:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[CMP25:%.*]] = icmp eq <8 x i16> [[TMP26]], [[TMP27]]
-// CHECK:   [[SEXT26:%.*]] = sext <8 x i1> [[CMP25]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT26]], ptr @bs, align 8
-// CHECK:   [[TMP28:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[CMP27:%.*]] = icmp eq <4 x i32> [[TMP28]], [[TMP29]]
-// CHECK:   [[SEXT28:%.*]] = sext <4 x i1> [[CMP27]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT28]], ptr @bi, align 8
-// CHECK:   [[TMP30:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP31:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[CMP29:%.*]] = icmp eq <4 x i32> [[TMP30]], [[TMP31]]
-// CHECK:   [[SEXT30:%.*]] = sext <4 x i1> [[CMP29]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT30]], ptr @bi, align 8
-// CHECK:   [[TMP32:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP33:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[CMP31:%.*]] = icmp eq <4 x i32> [[TMP32]], [[TMP33]]
-// CHECK:   [[SEXT32:%.*]] = sext <4 x i1> [[CMP31]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT32]], ptr @bi, align 8
-// CHECK:   [[TMP34:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP35:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[CMP33:%.*]] = icmp eq <4 x i32> [[TMP34]], [[TMP35]]
-// CHECK:   [[SEXT34:%.*]] = sext <4 x i1> [[CMP33]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT34]], ptr @bi, align 8
-// CHECK:   [[TMP36:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP37:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[CMP35:%.*]] = icmp eq <4 x i32> [[TMP36]], [[TMP37]]
-// CHECK:   [[SEXT36:%.*]] = sext <4 x i1> [[CMP35]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT36]], ptr @bi, align 8
-// CHECK:   [[TMP38:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP39:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[CMP37:%.*]] = icmp eq <4 x i32> [[TMP38]], [[TMP39]]
-// CHECK:   [[SEXT38:%.*]] = sext <4 x i1> [[CMP37]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT38]], ptr @bi, align 8
-// CHECK:   [[TMP40:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP41:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[CMP39:%.*]] = icmp eq <4 x i32> [[TMP40]], [[TMP41]]
-// CHECK:   [[SEXT40:%.*]] = sext <4 x i1> [[CMP39]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT40]], ptr @bi, align 8
-// CHECK:   [[TMP42:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP43:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[CMP41:%.*]] = icmp eq <2 x i64> [[TMP42]], [[TMP43]]
-// CHECK:   [[SEXT42:%.*]] = sext <2 x i1> [[CMP41]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT42]], ptr @bl, align 8
-// CHECK:   [[TMP44:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP45:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[CMP43:%.*]] = icmp eq <2 x i64> [[TMP44]], [[TMP45]]
-// CHECK:   [[SEXT44:%.*]] = sext <2 x i1> [[CMP43]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT44]], ptr @bl, align 8
-// CHECK:   [[TMP46:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP47:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[CMP45:%.*]] = icmp eq <2 x i64> [[TMP46]], [[TMP47]]
-// CHECK:   [[SEXT46:%.*]] = sext <2 x i1> [[CMP45]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT46]], ptr @bl, align 8
-// CHECK:   [[TMP48:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP49:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[CMP47:%.*]] = icmp eq <2 x i64> [[TMP48]], [[TMP49]]
-// CHECK:   [[SEXT48:%.*]] = sext <2 x i1> [[CMP47]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT48]], ptr @bl, align 8
-// CHECK:   [[TMP50:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP51:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[CMP49:%.*]] = icmp eq <2 x i64> [[TMP50]], [[TMP51]]
-// CHECK:   [[SEXT50:%.*]] = sext <2 x i1> [[CMP49]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT50]], ptr @bl, align 8
-// CHECK:   [[TMP52:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP53:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[CMP51:%.*]] = icmp eq <2 x i64> [[TMP52]], [[TMP53]]
-// CHECK:   [[SEXT52:%.*]] = sext <2 x i1> [[CMP51]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT52]], ptr @bl, align 8
-// CHECK:   [[TMP54:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP55:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[CMP53:%.*]] = icmp eq <2 x i64> [[TMP54]], [[TMP55]]
-// CHECK:   [[SEXT54:%.*]] = sext <2 x i1> [[CMP53]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT54]], ptr @bl, align 8
-// CHECK:   [[TMP56:%.*]] = load volatile <2 x double>, ptr @fd, align 8
-// CHECK:   [[TMP57:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[CMP55:%.*]] = fcmp oeq <2 x double> [[TMP56]], [[TMP57]]
-// CHECK:   [[SEXT56:%.*]] = sext <2 x i1> [[CMP55]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT56]], ptr @bl, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_cmpeq(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[CMP:%.*]] = icmp eq <16 x i8> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[CMP1:%.*]] = icmp eq <16 x i8> [[TMP2]], [[TMP3]]
+// CHECK-NEXT:    [[SEXT2:%.*]] = sext <16 x i1> [[CMP1]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT2]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[CMP3:%.*]] = icmp eq <16 x i8> [[TMP4]], [[TMP5]]
+// CHECK-NEXT:    [[SEXT4:%.*]] = sext <16 x i1> [[CMP3]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT4]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[CMP5:%.*]] = icmp eq <16 x i8> [[TMP6]], [[TMP7]]
+// CHECK-NEXT:    [[SEXT6:%.*]] = sext <16 x i1> [[CMP5]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT6]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[CMP7:%.*]] = icmp eq <16 x i8> [[TMP8]], [[TMP9]]
+// CHECK-NEXT:    [[SEXT8:%.*]] = sext <16 x i1> [[CMP7]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT8]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[CMP9:%.*]] = icmp eq <16 x i8> [[TMP10]], [[TMP11]]
+// CHECK-NEXT:    [[SEXT10:%.*]] = sext <16 x i1> [[CMP9]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT10]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[CMP11:%.*]] = icmp eq <16 x i8> [[TMP12]], [[TMP13]]
+// CHECK-NEXT:    [[SEXT12:%.*]] = sext <16 x i1> [[CMP11]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT12]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[CMP13:%.*]] = icmp eq <8 x i16> [[TMP14]], [[TMP15]]
+// CHECK-NEXT:    [[SEXT14:%.*]] = sext <8 x i1> [[CMP13]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT14]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[CMP15:%.*]] = icmp eq <8 x i16> [[TMP16]], [[TMP17]]
+// CHECK-NEXT:    [[SEXT16:%.*]] = sext <8 x i1> [[CMP15]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT16]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[CMP17:%.*]] = icmp eq <8 x i16> [[TMP18]], [[TMP19]]
+// CHECK-NEXT:    [[SEXT18:%.*]] = sext <8 x i1> [[CMP17]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT18]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[CMP19:%.*]] = icmp eq <8 x i16> [[TMP20]], [[TMP21]]
+// CHECK-NEXT:    [[SEXT20:%.*]] = sext <8 x i1> [[CMP19]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT20]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[CMP21:%.*]] = icmp eq <8 x i16> [[TMP22]], [[TMP23]]
+// CHECK-NEXT:    [[SEXT22:%.*]] = sext <8 x i1> [[CMP21]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT22]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[CMP23:%.*]] = icmp eq <8 x i16> [[TMP24]], [[TMP25]]
+// CHECK-NEXT:    [[SEXT24:%.*]] = sext <8 x i1> [[CMP23]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT24]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP26:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP27:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[CMP25:%.*]] = icmp eq <8 x i16> [[TMP26]], [[TMP27]]
+// CHECK-NEXT:    [[SEXT26:%.*]] = sext <8 x i1> [[CMP25]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT26]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP28:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[CMP27:%.*]] = icmp eq <4 x i32> [[TMP28]], [[TMP29]]
+// CHECK-NEXT:    [[SEXT28:%.*]] = sext <4 x i1> [[CMP27]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT28]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP30:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP31:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[CMP29:%.*]] = icmp eq <4 x i32> [[TMP30]], [[TMP31]]
+// CHECK-NEXT:    [[SEXT30:%.*]] = sext <4 x i1> [[CMP29]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT30]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP32:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP33:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[CMP31:%.*]] = icmp eq <4 x i32> [[TMP32]], [[TMP33]]
+// CHECK-NEXT:    [[SEXT32:%.*]] = sext <4 x i1> [[CMP31]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT32]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP34:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP35:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[CMP33:%.*]] = icmp eq <4 x i32> [[TMP34]], [[TMP35]]
+// CHECK-NEXT:    [[SEXT34:%.*]] = sext <4 x i1> [[CMP33]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT34]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP36:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP37:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[CMP35:%.*]] = icmp eq <4 x i32> [[TMP36]], [[TMP37]]
+// CHECK-NEXT:    [[SEXT36:%.*]] = sext <4 x i1> [[CMP35]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT36]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP38:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP39:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[CMP37:%.*]] = icmp eq <4 x i32> [[TMP38]], [[TMP39]]
+// CHECK-NEXT:    [[SEXT38:%.*]] = sext <4 x i1> [[CMP37]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT38]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP40:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP41:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[CMP39:%.*]] = icmp eq <4 x i32> [[TMP40]], [[TMP41]]
+// CHECK-NEXT:    [[SEXT40:%.*]] = sext <4 x i1> [[CMP39]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT40]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP42:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP43:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[CMP41:%.*]] = icmp eq <2 x i64> [[TMP42]], [[TMP43]]
+// CHECK-NEXT:    [[SEXT42:%.*]] = sext <2 x i1> [[CMP41]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT42]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP44:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP45:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[CMP43:%.*]] = icmp eq <2 x i64> [[TMP44]], [[TMP45]]
+// CHECK-NEXT:    [[SEXT44:%.*]] = sext <2 x i1> [[CMP43]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT44]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP46:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP47:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[CMP45:%.*]] = icmp eq <2 x i64> [[TMP46]], [[TMP47]]
+// CHECK-NEXT:    [[SEXT46:%.*]] = sext <2 x i1> [[CMP45]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT46]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP48:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP49:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[CMP47:%.*]] = icmp eq <2 x i64> [[TMP48]], [[TMP49]]
+// CHECK-NEXT:    [[SEXT48:%.*]] = sext <2 x i1> [[CMP47]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT48]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP50:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP51:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[CMP49:%.*]] = icmp eq <2 x i64> [[TMP50]], [[TMP51]]
+// CHECK-NEXT:    [[SEXT50:%.*]] = sext <2 x i1> [[CMP49]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT50]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP52:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP53:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[CMP51:%.*]] = icmp eq <2 x i64> [[TMP52]], [[TMP53]]
+// CHECK-NEXT:    [[SEXT52:%.*]] = sext <2 x i1> [[CMP51]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT52]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP54:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP55:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[CMP53:%.*]] = icmp eq <2 x i64> [[TMP54]], [[TMP55]]
+// CHECK-NEXT:    [[SEXT54:%.*]] = sext <2 x i1> [[CMP53]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT54]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP56:%.*]] = load volatile <2 x double>, ptr @fd, align 8
+// CHECK-NEXT:    [[TMP57:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[CMP55:%.*]] = fcmp oeq <2 x double> [[TMP56]], [[TMP57]]
+// CHECK-NEXT:    [[SEXT56:%.*]] = sext <2 x i1> [[CMP55]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT56]], ptr @bl, align 8
+// CHECK-NEXT:    ret void
+//
 void test_cmpeq(void) {
 
   bc = sc == sc2;
@@ -2832,153 +2920,156 @@ void test_cmpeq(void) {
   bl = fd == fd2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_cmpne() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[CMP:%.*]] = icmp ne <16 x i8> [[TMP0]], [[TMP1]]
-// CHECK:   [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT]], ptr @bc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[CMP1:%.*]] = icmp ne <16 x i8> [[TMP2]], [[TMP3]]
-// CHECK:   [[SEXT2:%.*]] = sext <16 x i1> [[CMP1]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT2]], ptr @bc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[CMP3:%.*]] = icmp ne <16 x i8> [[TMP4]], [[TMP5]]
-// CHECK:   [[SEXT4:%.*]] = sext <16 x i1> [[CMP3]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT4]], ptr @bc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[CMP5:%.*]] = icmp ne <16 x i8> [[TMP6]], [[TMP7]]
-// CHECK:   [[SEXT6:%.*]] = sext <16 x i1> [[CMP5]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT6]], ptr @bc, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[CMP7:%.*]] = icmp ne <16 x i8> [[TMP8]], [[TMP9]]
-// CHECK:   [[SEXT8:%.*]] = sext <16 x i1> [[CMP7]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT8]], ptr @bc, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[CMP9:%.*]] = icmp ne <16 x i8> [[TMP10]], [[TMP11]]
-// CHECK:   [[SEXT10:%.*]] = sext <16 x i1> [[CMP9]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT10]], ptr @bc, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[CMP11:%.*]] = icmp ne <16 x i8> [[TMP12]], [[TMP13]]
-// CHECK:   [[SEXT12:%.*]] = sext <16 x i1> [[CMP11]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT12]], ptr @bc, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[CMP13:%.*]] = icmp ne <8 x i16> [[TMP14]], [[TMP15]]
-// CHECK:   [[SEXT14:%.*]] = sext <8 x i1> [[CMP13]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT14]], ptr @bs, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[CMP15:%.*]] = icmp ne <8 x i16> [[TMP16]], [[TMP17]]
-// CHECK:   [[SEXT16:%.*]] = sext <8 x i1> [[CMP15]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT16]], ptr @bs, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP19:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[CMP17:%.*]] = icmp ne <8 x i16> [[TMP18]], [[TMP19]]
-// CHECK:   [[SEXT18:%.*]] = sext <8 x i1> [[CMP17]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT18]], ptr @bs, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[CMP19:%.*]] = icmp ne <8 x i16> [[TMP20]], [[TMP21]]
-// CHECK:   [[SEXT20:%.*]] = sext <8 x i1> [[CMP19]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT20]], ptr @bs, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[CMP21:%.*]] = icmp ne <8 x i16> [[TMP22]], [[TMP23]]
-// CHECK:   [[SEXT22:%.*]] = sext <8 x i1> [[CMP21]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT22]], ptr @bs, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[CMP23:%.*]] = icmp ne <8 x i16> [[TMP24]], [[TMP25]]
-// CHECK:   [[SEXT24:%.*]] = sext <8 x i1> [[CMP23]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT24]], ptr @bs, align 8
-// CHECK:   [[TMP26:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP27:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[CMP25:%.*]] = icmp ne <8 x i16> [[TMP26]], [[TMP27]]
-// CHECK:   [[SEXT26:%.*]] = sext <8 x i1> [[CMP25]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT26]], ptr @bs, align 8
-// CHECK:   [[TMP28:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[CMP27:%.*]] = icmp ne <4 x i32> [[TMP28]], [[TMP29]]
-// CHECK:   [[SEXT28:%.*]] = sext <4 x i1> [[CMP27]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT28]], ptr @bi, align 8
-// CHECK:   [[TMP30:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP31:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[CMP29:%.*]] = icmp ne <4 x i32> [[TMP30]], [[TMP31]]
-// CHECK:   [[SEXT30:%.*]] = sext <4 x i1> [[CMP29]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT30]], ptr @bi, align 8
-// CHECK:   [[TMP32:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP33:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[CMP31:%.*]] = icmp ne <4 x i32> [[TMP32]], [[TMP33]]
-// CHECK:   [[SEXT32:%.*]] = sext <4 x i1> [[CMP31]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT32]], ptr @bi, align 8
-// CHECK:   [[TMP34:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP35:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[CMP33:%.*]] = icmp ne <4 x i32> [[TMP34]], [[TMP35]]
-// CHECK:   [[SEXT34:%.*]] = sext <4 x i1> [[CMP33]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT34]], ptr @bi, align 8
-// CHECK:   [[TMP36:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP37:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[CMP35:%.*]] = icmp ne <4 x i32> [[TMP36]], [[TMP37]]
-// CHECK:   [[SEXT36:%.*]] = sext <4 x i1> [[CMP35]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT36]], ptr @bi, align 8
-// CHECK:   [[TMP38:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP39:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[CMP37:%.*]] = icmp ne <4 x i32> [[TMP38]], [[TMP39]]
-// CHECK:   [[SEXT38:%.*]] = sext <4 x i1> [[CMP37]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT38]], ptr @bi, align 8
-// CHECK:   [[TMP40:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP41:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[CMP39:%.*]] = icmp ne <4 x i32> [[TMP40]], [[TMP41]]
-// CHECK:   [[SEXT40:%.*]] = sext <4 x i1> [[CMP39]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT40]], ptr @bi, align 8
-// CHECK:   [[TMP42:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP43:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[CMP41:%.*]] = icmp ne <2 x i64> [[TMP42]], [[TMP43]]
-// CHECK:   [[SEXT42:%.*]] = sext <2 x i1> [[CMP41]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT42]], ptr @bl, align 8
-// CHECK:   [[TMP44:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP45:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[CMP43:%.*]] = icmp ne <2 x i64> [[TMP44]], [[TMP45]]
-// CHECK:   [[SEXT44:%.*]] = sext <2 x i1> [[CMP43]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT44]], ptr @bl, align 8
-// CHECK:   [[TMP46:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP47:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[CMP45:%.*]] = icmp ne <2 x i64> [[TMP46]], [[TMP47]]
-// CHECK:   [[SEXT46:%.*]] = sext <2 x i1> [[CMP45]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT46]], ptr @bl, align 8
-// CHECK:   [[TMP48:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP49:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[CMP47:%.*]] = icmp ne <2 x i64> [[TMP48]], [[TMP49]]
-// CHECK:   [[SEXT48:%.*]] = sext <2 x i1> [[CMP47]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT48]], ptr @bl, align 8
-// CHECK:   [[TMP50:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP51:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[CMP49:%.*]] = icmp ne <2 x i64> [[TMP50]], [[TMP51]]
-// CHECK:   [[SEXT50:%.*]] = sext <2 x i1> [[CMP49]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT50]], ptr @bl, align 8
-// CHECK:   [[TMP52:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP53:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[CMP51:%.*]] = icmp ne <2 x i64> [[TMP52]], [[TMP53]]
-// CHECK:   [[SEXT52:%.*]] = sext <2 x i1> [[CMP51]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT52]], ptr @bl, align 8
-// CHECK:   [[TMP54:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP55:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[CMP53:%.*]] = icmp ne <2 x i64> [[TMP54]], [[TMP55]]
-// CHECK:   [[SEXT54:%.*]] = sext <2 x i1> [[CMP53]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT54]], ptr @bl, align 8
-// CHECK:   [[TMP56:%.*]] = load volatile <2 x double>, ptr @fd, align 8
-// CHECK:   [[TMP57:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[CMP55:%.*]] = fcmp une <2 x double> [[TMP56]], [[TMP57]]
-// CHECK:   [[SEXT56:%.*]] = sext <2 x i1> [[CMP55]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT56]], ptr @bl, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_cmpne(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[CMP:%.*]] = icmp ne <16 x i8> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[CMP1:%.*]] = icmp ne <16 x i8> [[TMP2]], [[TMP3]]
+// CHECK-NEXT:    [[SEXT2:%.*]] = sext <16 x i1> [[CMP1]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT2]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[CMP3:%.*]] = icmp ne <16 x i8> [[TMP4]], [[TMP5]]
+// CHECK-NEXT:    [[SEXT4:%.*]] = sext <16 x i1> [[CMP3]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT4]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[CMP5:%.*]] = icmp ne <16 x i8> [[TMP6]], [[TMP7]]
+// CHECK-NEXT:    [[SEXT6:%.*]] = sext <16 x i1> [[CMP5]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT6]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[CMP7:%.*]] = icmp ne <16 x i8> [[TMP8]], [[TMP9]]
+// CHECK-NEXT:    [[SEXT8:%.*]] = sext <16 x i1> [[CMP7]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT8]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[CMP9:%.*]] = icmp ne <16 x i8> [[TMP10]], [[TMP11]]
+// CHECK-NEXT:    [[SEXT10:%.*]] = sext <16 x i1> [[CMP9]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT10]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[CMP11:%.*]] = icmp ne <16 x i8> [[TMP12]], [[TMP13]]
+// CHECK-NEXT:    [[SEXT12:%.*]] = sext <16 x i1> [[CMP11]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT12]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[CMP13:%.*]] = icmp ne <8 x i16> [[TMP14]], [[TMP15]]
+// CHECK-NEXT:    [[SEXT14:%.*]] = sext <8 x i1> [[CMP13]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT14]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[CMP15:%.*]] = icmp ne <8 x i16> [[TMP16]], [[TMP17]]
+// CHECK-NEXT:    [[SEXT16:%.*]] = sext <8 x i1> [[CMP15]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT16]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[CMP17:%.*]] = icmp ne <8 x i16> [[TMP18]], [[TMP19]]
+// CHECK-NEXT:    [[SEXT18:%.*]] = sext <8 x i1> [[CMP17]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT18]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[CMP19:%.*]] = icmp ne <8 x i16> [[TMP20]], [[TMP21]]
+// CHECK-NEXT:    [[SEXT20:%.*]] = sext <8 x i1> [[CMP19]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT20]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[CMP21:%.*]] = icmp ne <8 x i16> [[TMP22]], [[TMP23]]
+// CHECK-NEXT:    [[SEXT22:%.*]] = sext <8 x i1> [[CMP21]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT22]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[CMP23:%.*]] = icmp ne <8 x i16> [[TMP24]], [[TMP25]]
+// CHECK-NEXT:    [[SEXT24:%.*]] = sext <8 x i1> [[CMP23]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT24]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP26:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP27:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[CMP25:%.*]] = icmp ne <8 x i16> [[TMP26]], [[TMP27]]
+// CHECK-NEXT:    [[SEXT26:%.*]] = sext <8 x i1> [[CMP25]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT26]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP28:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP29:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[CMP27:%.*]] = icmp ne <4 x i32> [[TMP28]], [[TMP29]]
+// CHECK-NEXT:    [[SEXT28:%.*]] = sext <4 x i1> [[CMP27]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT28]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP30:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP31:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[CMP29:%.*]] = icmp ne <4 x i32> [[TMP30]], [[TMP31]]
+// CHECK-NEXT:    [[SEXT30:%.*]] = sext <4 x i1> [[CMP29]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT30]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP32:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP33:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[CMP31:%.*]] = icmp ne <4 x i32> [[TMP32]], [[TMP33]]
+// CHECK-NEXT:    [[SEXT32:%.*]] = sext <4 x i1> [[CMP31]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT32]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP34:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP35:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[CMP33:%.*]] = icmp ne <4 x i32> [[TMP34]], [[TMP35]]
+// CHECK-NEXT:    [[SEXT34:%.*]] = sext <4 x i1> [[CMP33]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT34]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP36:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP37:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[CMP35:%.*]] = icmp ne <4 x i32> [[TMP36]], [[TMP37]]
+// CHECK-NEXT:    [[SEXT36:%.*]] = sext <4 x i1> [[CMP35]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT36]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP38:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP39:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[CMP37:%.*]] = icmp ne <4 x i32> [[TMP38]], [[TMP39]]
+// CHECK-NEXT:    [[SEXT38:%.*]] = sext <4 x i1> [[CMP37]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT38]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP40:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP41:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[CMP39:%.*]] = icmp ne <4 x i32> [[TMP40]], [[TMP41]]
+// CHECK-NEXT:    [[SEXT40:%.*]] = sext <4 x i1> [[CMP39]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT40]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP42:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP43:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[CMP41:%.*]] = icmp ne <2 x i64> [[TMP42]], [[TMP43]]
+// CHECK-NEXT:    [[SEXT42:%.*]] = sext <2 x i1> [[CMP41]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT42]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP44:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP45:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[CMP43:%.*]] = icmp ne <2 x i64> [[TMP44]], [[TMP45]]
+// CHECK-NEXT:    [[SEXT44:%.*]] = sext <2 x i1> [[CMP43]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT44]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP46:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP47:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[CMP45:%.*]] = icmp ne <2 x i64> [[TMP46]], [[TMP47]]
+// CHECK-NEXT:    [[SEXT46:%.*]] = sext <2 x i1> [[CMP45]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT46]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP48:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP49:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[CMP47:%.*]] = icmp ne <2 x i64> [[TMP48]], [[TMP49]]
+// CHECK-NEXT:    [[SEXT48:%.*]] = sext <2 x i1> [[CMP47]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT48]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP50:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP51:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[CMP49:%.*]] = icmp ne <2 x i64> [[TMP50]], [[TMP51]]
+// CHECK-NEXT:    [[SEXT50:%.*]] = sext <2 x i1> [[CMP49]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT50]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP52:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP53:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[CMP51:%.*]] = icmp ne <2 x i64> [[TMP52]], [[TMP53]]
+// CHECK-NEXT:    [[SEXT52:%.*]] = sext <2 x i1> [[CMP51]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT52]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP54:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP55:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[CMP53:%.*]] = icmp ne <2 x i64> [[TMP54]], [[TMP55]]
+// CHECK-NEXT:    [[SEXT54:%.*]] = sext <2 x i1> [[CMP53]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT54]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP56:%.*]] = load volatile <2 x double>, ptr @fd, align 8
+// CHECK-NEXT:    [[TMP57:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[CMP55:%.*]] = fcmp une <2 x double> [[TMP56]], [[TMP57]]
+// CHECK-NEXT:    [[SEXT56:%.*]] = sext <2 x i1> [[CMP55]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT56]], ptr @bl, align 8
+// CHECK-NEXT:    ret void
+//
 void test_cmpne(void) {
 
   bc = sc != sc2;
@@ -3016,73 +3107,76 @@ void test_cmpne(void) {
   bl = fd != fd2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_cmpge() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[CMP:%.*]] = icmp sge <16 x i8> [[TMP0]], [[TMP1]]
-// CHECK:   [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT]], ptr @bc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[CMP1:%.*]] = icmp uge <16 x i8> [[TMP2]], [[TMP3]]
-// CHECK:   [[SEXT2:%.*]] = sext <16 x i1> [[CMP1]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT2]], ptr @bc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[CMP3:%.*]] = icmp uge <16 x i8> [[TMP4]], [[TMP5]]
-// CHECK:   [[SEXT4:%.*]] = sext <16 x i1> [[CMP3]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT4]], ptr @bc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[CMP5:%.*]] = icmp sge <8 x i16> [[TMP6]], [[TMP7]]
-// CHECK:   [[SEXT6:%.*]] = sext <8 x i1> [[CMP5]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT6]], ptr @bs, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[CMP7:%.*]] = icmp uge <8 x i16> [[TMP8]], [[TMP9]]
-// CHECK:   [[SEXT8:%.*]] = sext <8 x i1> [[CMP7]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT8]], ptr @bs, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[CMP9:%.*]] = icmp uge <8 x i16> [[TMP10]], [[TMP11]]
-// CHECK:   [[SEXT10:%.*]] = sext <8 x i1> [[CMP9]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT10]], ptr @bs, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[CMP11:%.*]] = icmp sge <4 x i32> [[TMP12]], [[TMP13]]
-// CHECK:   [[SEXT12:%.*]] = sext <4 x i1> [[CMP11]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT12]], ptr @bi, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[CMP13:%.*]] = icmp uge <4 x i32> [[TMP14]], [[TMP15]]
-// CHECK:   [[SEXT14:%.*]] = sext <4 x i1> [[CMP13]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT14]], ptr @bi, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[CMP15:%.*]] = icmp uge <4 x i32> [[TMP16]], [[TMP17]]
-// CHECK:   [[SEXT16:%.*]] = sext <4 x i1> [[CMP15]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT16]], ptr @bi, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP19:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[CMP17:%.*]] = icmp sge <2 x i64> [[TMP18]], [[TMP19]]
-// CHECK:   [[SEXT18:%.*]] = sext <2 x i1> [[CMP17]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT18]], ptr @bl, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[CMP19:%.*]] = icmp uge <2 x i64> [[TMP20]], [[TMP21]]
-// CHECK:   [[SEXT20:%.*]] = sext <2 x i1> [[CMP19]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT20]], ptr @bl, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[CMP21:%.*]] = icmp uge <2 x i64> [[TMP22]], [[TMP23]]
-// CHECK:   [[SEXT22:%.*]] = sext <2 x i1> [[CMP21]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT22]], ptr @bl, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <2 x double>, ptr @fd, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[CMP23:%.*]] = fcmp oge <2 x double> [[TMP24]], [[TMP25]]
-// CHECK:   [[SEXT24:%.*]] = sext <2 x i1> [[CMP23]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT24]], ptr @bl, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_cmpge(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[CMP:%.*]] = icmp sge <16 x i8> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[CMP1:%.*]] = icmp uge <16 x i8> [[TMP2]], [[TMP3]]
+// CHECK-NEXT:    [[SEXT2:%.*]] = sext <16 x i1> [[CMP1]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT2]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[CMP3:%.*]] = icmp uge <16 x i8> [[TMP4]], [[TMP5]]
+// CHECK-NEXT:    [[SEXT4:%.*]] = sext <16 x i1> [[CMP3]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT4]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[CMP5:%.*]] = icmp sge <8 x i16> [[TMP6]], [[TMP7]]
+// CHECK-NEXT:    [[SEXT6:%.*]] = sext <8 x i1> [[CMP5]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT6]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[CMP7:%.*]] = icmp uge <8 x i16> [[TMP8]], [[TMP9]]
+// CHECK-NEXT:    [[SEXT8:%.*]] = sext <8 x i1> [[CMP7]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT8]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[CMP9:%.*]] = icmp uge <8 x i16> [[TMP10]], [[TMP11]]
+// CHECK-NEXT:    [[SEXT10:%.*]] = sext <8 x i1> [[CMP9]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT10]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[CMP11:%.*]] = icmp sge <4 x i32> [[TMP12]], [[TMP13]]
+// CHECK-NEXT:    [[SEXT12:%.*]] = sext <4 x i1> [[CMP11]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT12]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[CMP13:%.*]] = icmp uge <4 x i32> [[TMP14]], [[TMP15]]
+// CHECK-NEXT:    [[SEXT14:%.*]] = sext <4 x i1> [[CMP13]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT14]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[CMP15:%.*]] = icmp uge <4 x i32> [[TMP16]], [[TMP17]]
+// CHECK-NEXT:    [[SEXT16:%.*]] = sext <4 x i1> [[CMP15]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT16]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[CMP17:%.*]] = icmp sge <2 x i64> [[TMP18]], [[TMP19]]
+// CHECK-NEXT:    [[SEXT18:%.*]] = sext <2 x i1> [[CMP17]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT18]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[CMP19:%.*]] = icmp uge <2 x i64> [[TMP20]], [[TMP21]]
+// CHECK-NEXT:    [[SEXT20:%.*]] = sext <2 x i1> [[CMP19]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT20]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[CMP21:%.*]] = icmp uge <2 x i64> [[TMP22]], [[TMP23]]
+// CHECK-NEXT:    [[SEXT22:%.*]] = sext <2 x i1> [[CMP21]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT22]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <2 x double>, ptr @fd, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[CMP23:%.*]] = fcmp oge <2 x double> [[TMP24]], [[TMP25]]
+// CHECK-NEXT:    [[SEXT24:%.*]] = sext <2 x i1> [[CMP23]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT24]], ptr @bl, align 8
+// CHECK-NEXT:    ret void
+//
 void test_cmpge(void) {
 
   bc = sc >= sc2;
@@ -3104,73 +3198,76 @@ void test_cmpge(void) {
   bl = fd >= fd2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_cmpgt() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[CMP:%.*]] = icmp sgt <16 x i8> [[TMP0]], [[TMP1]]
-// CHECK:   [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT]], ptr @bc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[CMP1:%.*]] = icmp ugt <16 x i8> [[TMP2]], [[TMP3]]
-// CHECK:   [[SEXT2:%.*]] = sext <16 x i1> [[CMP1]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT2]], ptr @bc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[CMP3:%.*]] = icmp ugt <16 x i8> [[TMP4]], [[TMP5]]
-// CHECK:   [[SEXT4:%.*]] = sext <16 x i1> [[CMP3]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT4]], ptr @bc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[CMP5:%.*]] = icmp sgt <8 x i16> [[TMP6]], [[TMP7]]
-// CHECK:   [[SEXT6:%.*]] = sext <8 x i1> [[CMP5]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT6]], ptr @bs, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[CMP7:%.*]] = icmp ugt <8 x i16> [[TMP8]], [[TMP9]]
-// CHECK:   [[SEXT8:%.*]] = sext <8 x i1> [[CMP7]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT8]], ptr @bs, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[CMP9:%.*]] = icmp ugt <8 x i16> [[TMP10]], [[TMP11]]
-// CHECK:   [[SEXT10:%.*]] = sext <8 x i1> [[CMP9]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT10]], ptr @bs, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[CMP11:%.*]] = icmp sgt <4 x i32> [[TMP12]], [[TMP13]]
-// CHECK:   [[SEXT12:%.*]] = sext <4 x i1> [[CMP11]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT12]], ptr @bi, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[CMP13:%.*]] = icmp ugt <4 x i32> [[TMP14]], [[TMP15]]
-// CHECK:   [[SEXT14:%.*]] = sext <4 x i1> [[CMP13]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT14]], ptr @bi, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[CMP15:%.*]] = icmp ugt <4 x i32> [[TMP16]], [[TMP17]]
-// CHECK:   [[SEXT16:%.*]] = sext <4 x i1> [[CMP15]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT16]], ptr @bi, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP19:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[CMP17:%.*]] = icmp sgt <2 x i64> [[TMP18]], [[TMP19]]
-// CHECK:   [[SEXT18:%.*]] = sext <2 x i1> [[CMP17]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT18]], ptr @bl, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[CMP19:%.*]] = icmp ugt <2 x i64> [[TMP20]], [[TMP21]]
-// CHECK:   [[SEXT20:%.*]] = sext <2 x i1> [[CMP19]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT20]], ptr @bl, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[CMP21:%.*]] = icmp ugt <2 x i64> [[TMP22]], [[TMP23]]
-// CHECK:   [[SEXT22:%.*]] = sext <2 x i1> [[CMP21]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT22]], ptr @bl, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <2 x double>, ptr @fd, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[CMP23:%.*]] = fcmp ogt <2 x double> [[TMP24]], [[TMP25]]
-// CHECK:   [[SEXT24:%.*]] = sext <2 x i1> [[CMP23]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT24]], ptr @bl, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_cmpgt(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[CMP:%.*]] = icmp sgt <16 x i8> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[CMP1:%.*]] = icmp ugt <16 x i8> [[TMP2]], [[TMP3]]
+// CHECK-NEXT:    [[SEXT2:%.*]] = sext <16 x i1> [[CMP1]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT2]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[CMP3:%.*]] = icmp ugt <16 x i8> [[TMP4]], [[TMP5]]
+// CHECK-NEXT:    [[SEXT4:%.*]] = sext <16 x i1> [[CMP3]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT4]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[CMP5:%.*]] = icmp sgt <8 x i16> [[TMP6]], [[TMP7]]
+// CHECK-NEXT:    [[SEXT6:%.*]] = sext <8 x i1> [[CMP5]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT6]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[CMP7:%.*]] = icmp ugt <8 x i16> [[TMP8]], [[TMP9]]
+// CHECK-NEXT:    [[SEXT8:%.*]] = sext <8 x i1> [[CMP7]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT8]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[CMP9:%.*]] = icmp ugt <8 x i16> [[TMP10]], [[TMP11]]
+// CHECK-NEXT:    [[SEXT10:%.*]] = sext <8 x i1> [[CMP9]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT10]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[CMP11:%.*]] = icmp sgt <4 x i32> [[TMP12]], [[TMP13]]
+// CHECK-NEXT:    [[SEXT12:%.*]] = sext <4 x i1> [[CMP11]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT12]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[CMP13:%.*]] = icmp ugt <4 x i32> [[TMP14]], [[TMP15]]
+// CHECK-NEXT:    [[SEXT14:%.*]] = sext <4 x i1> [[CMP13]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT14]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[CMP15:%.*]] = icmp ugt <4 x i32> [[TMP16]], [[TMP17]]
+// CHECK-NEXT:    [[SEXT16:%.*]] = sext <4 x i1> [[CMP15]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT16]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[CMP17:%.*]] = icmp sgt <2 x i64> [[TMP18]], [[TMP19]]
+// CHECK-NEXT:    [[SEXT18:%.*]] = sext <2 x i1> [[CMP17]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT18]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[CMP19:%.*]] = icmp ugt <2 x i64> [[TMP20]], [[TMP21]]
+// CHECK-NEXT:    [[SEXT20:%.*]] = sext <2 x i1> [[CMP19]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT20]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[CMP21:%.*]] = icmp ugt <2 x i64> [[TMP22]], [[TMP23]]
+// CHECK-NEXT:    [[SEXT22:%.*]] = sext <2 x i1> [[CMP21]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT22]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <2 x double>, ptr @fd, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[CMP23:%.*]] = fcmp ogt <2 x double> [[TMP24]], [[TMP25]]
+// CHECK-NEXT:    [[SEXT24:%.*]] = sext <2 x i1> [[CMP23]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT24]], ptr @bl, align 8
+// CHECK-NEXT:    ret void
+//
 void test_cmpgt(void) {
 
   bc = sc > sc2;
@@ -3192,73 +3289,76 @@ void test_cmpgt(void) {
   bl = fd > fd2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_cmple() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[CMP:%.*]] = icmp sle <16 x i8> [[TMP0]], [[TMP1]]
-// CHECK:   [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT]], ptr @bc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[CMP1:%.*]] = icmp ule <16 x i8> [[TMP2]], [[TMP3]]
-// CHECK:   [[SEXT2:%.*]] = sext <16 x i1> [[CMP1]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT2]], ptr @bc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[CMP3:%.*]] = icmp ule <16 x i8> [[TMP4]], [[TMP5]]
-// CHECK:   [[SEXT4:%.*]] = sext <16 x i1> [[CMP3]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT4]], ptr @bc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[CMP5:%.*]] = icmp sle <8 x i16> [[TMP6]], [[TMP7]]
-// CHECK:   [[SEXT6:%.*]] = sext <8 x i1> [[CMP5]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT6]], ptr @bs, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[CMP7:%.*]] = icmp ule <8 x i16> [[TMP8]], [[TMP9]]
-// CHECK:   [[SEXT8:%.*]] = sext <8 x i1> [[CMP7]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT8]], ptr @bs, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[CMP9:%.*]] = icmp ule <8 x i16> [[TMP10]], [[TMP11]]
-// CHECK:   [[SEXT10:%.*]] = sext <8 x i1> [[CMP9]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT10]], ptr @bs, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[CMP11:%.*]] = icmp sle <4 x i32> [[TMP12]], [[TMP13]]
-// CHECK:   [[SEXT12:%.*]] = sext <4 x i1> [[CMP11]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT12]], ptr @bi, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[CMP13:%.*]] = icmp ule <4 x i32> [[TMP14]], [[TMP15]]
-// CHECK:   [[SEXT14:%.*]] = sext <4 x i1> [[CMP13]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT14]], ptr @bi, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[CMP15:%.*]] = icmp ule <4 x i32> [[TMP16]], [[TMP17]]
-// CHECK:   [[SEXT16:%.*]] = sext <4 x i1> [[CMP15]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT16]], ptr @bi, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP19:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[CMP17:%.*]] = icmp sle <2 x i64> [[TMP18]], [[TMP19]]
-// CHECK:   [[SEXT18:%.*]] = sext <2 x i1> [[CMP17]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT18]], ptr @bl, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[CMP19:%.*]] = icmp ule <2 x i64> [[TMP20]], [[TMP21]]
-// CHECK:   [[SEXT20:%.*]] = sext <2 x i1> [[CMP19]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT20]], ptr @bl, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[CMP21:%.*]] = icmp ule <2 x i64> [[TMP22]], [[TMP23]]
-// CHECK:   [[SEXT22:%.*]] = sext <2 x i1> [[CMP21]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT22]], ptr @bl, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <2 x double>, ptr @fd, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[CMP23:%.*]] = fcmp ole <2 x double> [[TMP24]], [[TMP25]]
-// CHECK:   [[SEXT24:%.*]] = sext <2 x i1> [[CMP23]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT24]], ptr @bl, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_cmple(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[CMP:%.*]] = icmp sle <16 x i8> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[CMP1:%.*]] = icmp ule <16 x i8> [[TMP2]], [[TMP3]]
+// CHECK-NEXT:    [[SEXT2:%.*]] = sext <16 x i1> [[CMP1]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT2]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[CMP3:%.*]] = icmp ule <16 x i8> [[TMP4]], [[TMP5]]
+// CHECK-NEXT:    [[SEXT4:%.*]] = sext <16 x i1> [[CMP3]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT4]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[CMP5:%.*]] = icmp sle <8 x i16> [[TMP6]], [[TMP7]]
+// CHECK-NEXT:    [[SEXT6:%.*]] = sext <8 x i1> [[CMP5]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT6]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[CMP7:%.*]] = icmp ule <8 x i16> [[TMP8]], [[TMP9]]
+// CHECK-NEXT:    [[SEXT8:%.*]] = sext <8 x i1> [[CMP7]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT8]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[CMP9:%.*]] = icmp ule <8 x i16> [[TMP10]], [[TMP11]]
+// CHECK-NEXT:    [[SEXT10:%.*]] = sext <8 x i1> [[CMP9]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT10]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[CMP11:%.*]] = icmp sle <4 x i32> [[TMP12]], [[TMP13]]
+// CHECK-NEXT:    [[SEXT12:%.*]] = sext <4 x i1> [[CMP11]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT12]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[CMP13:%.*]] = icmp ule <4 x i32> [[TMP14]], [[TMP15]]
+// CHECK-NEXT:    [[SEXT14:%.*]] = sext <4 x i1> [[CMP13]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT14]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[CMP15:%.*]] = icmp ule <4 x i32> [[TMP16]], [[TMP17]]
+// CHECK-NEXT:    [[SEXT16:%.*]] = sext <4 x i1> [[CMP15]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT16]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[CMP17:%.*]] = icmp sle <2 x i64> [[TMP18]], [[TMP19]]
+// CHECK-NEXT:    [[SEXT18:%.*]] = sext <2 x i1> [[CMP17]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT18]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[CMP19:%.*]] = icmp ule <2 x i64> [[TMP20]], [[TMP21]]
+// CHECK-NEXT:    [[SEXT20:%.*]] = sext <2 x i1> [[CMP19]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT20]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[CMP21:%.*]] = icmp ule <2 x i64> [[TMP22]], [[TMP23]]
+// CHECK-NEXT:    [[SEXT22:%.*]] = sext <2 x i1> [[CMP21]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT22]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <2 x double>, ptr @fd, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[CMP23:%.*]] = fcmp ole <2 x double> [[TMP24]], [[TMP25]]
+// CHECK-NEXT:    [[SEXT24:%.*]] = sext <2 x i1> [[CMP23]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT24]], ptr @bl, align 8
+// CHECK-NEXT:    ret void
+//
 void test_cmple(void) {
 
   bc = sc <= sc2;
@@ -3280,73 +3380,76 @@ void test_cmple(void) {
   bl = fd <= fd2;
 }
 
-// CHECK-LABEL: define{{.*}} void @test_cmplt() #0 {
-// CHECK:   [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
-// CHECK:   [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
-// CHECK:   [[CMP:%.*]] = icmp slt <16 x i8> [[TMP0]], [[TMP1]]
-// CHECK:   [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT]], ptr @bc, align 8
-// CHECK:   [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
-// CHECK:   [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
-// CHECK:   [[CMP1:%.*]] = icmp ult <16 x i8> [[TMP2]], [[TMP3]]
-// CHECK:   [[SEXT2:%.*]] = sext <16 x i1> [[CMP1]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT2]], ptr @bc, align 8
-// CHECK:   [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
-// CHECK:   [[TMP5:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
-// CHECK:   [[CMP3:%.*]] = icmp ult <16 x i8> [[TMP4]], [[TMP5]]
-// CHECK:   [[SEXT4:%.*]] = sext <16 x i1> [[CMP3]] to <16 x i8>
-// CHECK:   store volatile <16 x i8> [[SEXT4]], ptr @bc, align 8
-// CHECK:   [[TMP6:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
-// CHECK:   [[TMP7:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
-// CHECK:   [[CMP5:%.*]] = icmp slt <8 x i16> [[TMP6]], [[TMP7]]
-// CHECK:   [[SEXT6:%.*]] = sext <8 x i1> [[CMP5]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT6]], ptr @bs, align 8
-// CHECK:   [[TMP8:%.*]] = load volatile <8 x i16>, ptr @us, align 8
-// CHECK:   [[TMP9:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
-// CHECK:   [[CMP7:%.*]] = icmp ult <8 x i16> [[TMP8]], [[TMP9]]
-// CHECK:   [[SEXT8:%.*]] = sext <8 x i1> [[CMP7]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT8]], ptr @bs, align 8
-// CHECK:   [[TMP10:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
-// CHECK:   [[TMP11:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
-// CHECK:   [[CMP9:%.*]] = icmp ult <8 x i16> [[TMP10]], [[TMP11]]
-// CHECK:   [[SEXT10:%.*]] = sext <8 x i1> [[CMP9]] to <8 x i16>
-// CHECK:   store volatile <8 x i16> [[SEXT10]], ptr @bs, align 8
-// CHECK:   [[TMP12:%.*]] = load volatile <4 x i32>, ptr @si, align 8
-// CHECK:   [[TMP13:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
-// CHECK:   [[CMP11:%.*]] = icmp slt <4 x i32> [[TMP12]], [[TMP13]]
-// CHECK:   [[SEXT12:%.*]] = sext <4 x i1> [[CMP11]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT12]], ptr @bi, align 8
-// CHECK:   [[TMP14:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
-// CHECK:   [[TMP15:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
-// CHECK:   [[CMP13:%.*]] = icmp ult <4 x i32> [[TMP14]], [[TMP15]]
-// CHECK:   [[SEXT14:%.*]] = sext <4 x i1> [[CMP13]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT14]], ptr @bi, align 8
-// CHECK:   [[TMP16:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
-// CHECK:   [[TMP17:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
-// CHECK:   [[CMP15:%.*]] = icmp ult <4 x i32> [[TMP16]], [[TMP17]]
-// CHECK:   [[SEXT16:%.*]] = sext <4 x i1> [[CMP15]] to <4 x i32>
-// CHECK:   store volatile <4 x i32> [[SEXT16]], ptr @bi, align 8
-// CHECK:   [[TMP18:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
-// CHECK:   [[TMP19:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
-// CHECK:   [[CMP17:%.*]] = icmp slt <2 x i64> [[TMP18]], [[TMP19]]
-// CHECK:   [[SEXT18:%.*]] = sext <2 x i1> [[CMP17]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT18]], ptr @bl, align 8
-// CHECK:   [[TMP20:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
-// CHECK:   [[TMP21:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
-// CHECK:   [[CMP19:%.*]] = icmp ult <2 x i64> [[TMP20]], [[TMP21]]
-// CHECK:   [[SEXT20:%.*]] = sext <2 x i1> [[CMP19]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT20]], ptr @bl, align 8
-// CHECK:   [[TMP22:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
-// CHECK:   [[TMP23:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
-// CHECK:   [[CMP21:%.*]] = icmp ult <2 x i64> [[TMP22]], [[TMP23]]
-// CHECK:   [[SEXT22:%.*]] = sext <2 x i1> [[CMP21]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT22]], ptr @bl, align 8
-// CHECK:   [[TMP24:%.*]] = load volatile <2 x double>, ptr @fd, align 8
-// CHECK:   [[TMP25:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
-// CHECK:   [[CMP23:%.*]] = fcmp olt <2 x double> [[TMP24]], [[TMP25]]
-// CHECK:   [[SEXT24:%.*]] = sext <2 x i1> [[CMP23]] to <2 x i64>
-// CHECK:   store volatile <2 x i64> [[SEXT24]], ptr @bl, align 8
-// CHECK:   ret void
+// CHECK-LABEL: define dso_local void @test_cmplt(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <16 x i8>, ptr @sc, align 8
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <16 x i8>, ptr @sc2, align 8
+// CHECK-NEXT:    [[CMP:%.*]] = icmp slt <16 x i8> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    [[SEXT:%.*]] = sext <16 x i1> [[CMP]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP2:%.*]] = load volatile <16 x i8>, ptr @uc, align 8
+// CHECK-NEXT:    [[TMP3:%.*]] = load volatile <16 x i8>, ptr @uc2, align 8
+// CHECK-NEXT:    [[CMP1:%.*]] = icmp ult <16 x i8> [[TMP2]], [[TMP3]]
+// CHECK-NEXT:    [[SEXT2:%.*]] = sext <16 x i1> [[CMP1]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT2]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP4:%.*]] = load volatile <16 x i8>, ptr @bc, align 8
+// CHECK-NEXT:    [[TMP5:%.*]] = load volatile <16 x i8>, ptr @bc2, align 8
+// CHECK-NEXT:    [[CMP3:%.*]] = icmp ult <16 x i8> [[TMP4]], [[TMP5]]
+// CHECK-NEXT:    [[SEXT4:%.*]] = sext <16 x i1> [[CMP3]] to <16 x i8>
+// CHECK-NEXT:    store volatile <16 x i8> [[SEXT4]], ptr @bc, align 8
+// CHECK-NEXT:    [[TMP6:%.*]] = load volatile <8 x i16>, ptr @ss, align 8
+// CHECK-NEXT:    [[TMP7:%.*]] = load volatile <8 x i16>, ptr @ss2, align 8
+// CHECK-NEXT:    [[CMP5:%.*]] = icmp slt <8 x i16> [[TMP6]], [[TMP7]]
+// CHECK-NEXT:    [[SEXT6:%.*]] = sext <8 x i1> [[CMP5]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT6]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP8:%.*]] = load volatile <8 x i16>, ptr @us, align 8
+// CHECK-NEXT:    [[TMP9:%.*]] = load volatile <8 x i16>, ptr @us2, align 8
+// CHECK-NEXT:    [[CMP7:%.*]] = icmp ult <8 x i16> [[TMP8]], [[TMP9]]
+// CHECK-NEXT:    [[SEXT8:%.*]] = sext <8 x i1> [[CMP7]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT8]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP10:%.*]] = load volatile <8 x i16>, ptr @bs, align 8
+// CHECK-NEXT:    [[TMP11:%.*]] = load volatile <8 x i16>, ptr @bs2, align 8
+// CHECK-NEXT:    [[CMP9:%.*]] = icmp ult <8 x i16> [[TMP10]], [[TMP11]]
+// CHECK-NEXT:    [[SEXT10:%.*]] = sext <8 x i1> [[CMP9]] to <8 x i16>
+// CHECK-NEXT:    store volatile <8 x i16> [[SEXT10]], ptr @bs, align 8
+// CHECK-NEXT:    [[TMP12:%.*]] = load volatile <4 x i32>, ptr @si, align 8
+// CHECK-NEXT:    [[TMP13:%.*]] = load volatile <4 x i32>, ptr @si2, align 8
+// CHECK-NEXT:    [[CMP11:%.*]] = icmp slt <4 x i32> [[TMP12]], [[TMP13]]
+// CHECK-NEXT:    [[SEXT12:%.*]] = sext <4 x i1> [[CMP11]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT12]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP14:%.*]] = load volatile <4 x i32>, ptr @ui, align 8
+// CHECK-NEXT:    [[TMP15:%.*]] = load volatile <4 x i32>, ptr @ui2, align 8
+// CHECK-NEXT:    [[CMP13:%.*]] = icmp ult <4 x i32> [[TMP14]], [[TMP15]]
+// CHECK-NEXT:    [[SEXT14:%.*]] = sext <4 x i1> [[CMP13]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT14]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP16:%.*]] = load volatile <4 x i32>, ptr @bi, align 8
+// CHECK-NEXT:    [[TMP17:%.*]] = load volatile <4 x i32>, ptr @bi2, align 8
+// CHECK-NEXT:    [[CMP15:%.*]] = icmp ult <4 x i32> [[TMP16]], [[TMP17]]
+// CHECK-NEXT:    [[SEXT16:%.*]] = sext <4 x i1> [[CMP15]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT16]], ptr @bi, align 8
+// CHECK-NEXT:    [[TMP18:%.*]] = load volatile <2 x i64>, ptr @sl, align 8
+// CHECK-NEXT:    [[TMP19:%.*]] = load volatile <2 x i64>, ptr @sl2, align 8
+// CHECK-NEXT:    [[CMP17:%.*]] = icmp slt <2 x i64> [[TMP18]], [[TMP19]]
+// CHECK-NEXT:    [[SEXT18:%.*]] = sext <2 x i1> [[CMP17]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT18]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP20:%.*]] = load volatile <2 x i64>, ptr @ul, align 8
+// CHECK-NEXT:    [[TMP21:%.*]] = load volatile <2 x i64>, ptr @ul2, align 8
+// CHECK-NEXT:    [[CMP19:%.*]] = icmp ult <2 x i64> [[TMP20]], [[TMP21]]
+// CHECK-NEXT:    [[SEXT20:%.*]] = sext <2 x i1> [[CMP19]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT20]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP22:%.*]] = load volatile <2 x i64>, ptr @bl, align 8
+// CHECK-NEXT:    [[TMP23:%.*]] = load volatile <2 x i64>, ptr @bl2, align 8
+// CHECK-NEXT:    [[CMP21:%.*]] = icmp ult <2 x i64> [[TMP22]], [[TMP23]]
+// CHECK-NEXT:    [[SEXT22:%.*]] = sext <2 x i1> [[CMP21]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT22]], ptr @bl, align 8
+// CHECK-NEXT:    [[TMP24:%.*]] = load volatile <2 x double>, ptr @fd, align 8
+// CHECK-NEXT:    [[TMP25:%.*]] = load volatile <2 x double>, ptr @fd2, align 8
+// CHECK-NEXT:    [[CMP23:%.*]] = fcmp olt <2 x double> [[TMP24]], [[TMP25]]
+// CHECK-NEXT:    [[SEXT24:%.*]] = sext <2 x i1> [[CMP23]] to <2 x i64>
+// CHECK-NEXT:    store volatile <2 x i64> [[SEXT24]], ptr @bl, align 8
+// CHECK-NEXT:    ret void
+//
 void test_cmplt(void) {
 
   bc = sc < sc2;

diff  --git a/clang/test/CodeGen/SystemZ/zvector2.c b/clang/test/CodeGen/SystemZ/zvector2.c
index 5b036433519bdd..b021ae85343536 100644
--- a/clang/test/CodeGen/SystemZ/zvector2.c
+++ b/clang/test/CodeGen/SystemZ/zvector2.c
@@ -1,194 +1,303 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
 // RUN: %clang_cc1 -triple s390x-linux-gnu -target-cpu z14 -fzvector \
 // RUN:  -O -emit-llvm -o - -W -Wall -Werror %s | FileCheck %s
 
 volatile vector float ff, ff2;
 volatile vector bool int bi;
 
+// CHECK-LABEL: define dso_local void @test_assign(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0:[0-9]+]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3:![0-9]+]]
+// CHECK-NEXT:    store volatile <4 x float> [[TMP0]], ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_assign (void)
 {
-// CHECK-LABEL: test_assign
-// CHECK: [[VAL:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: store volatile <4 x float> [[VAL]], ptr @ff
   ff = ff2;
 }
 
+// CHECK-LABEL: define dso_local void @test_pos(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    store volatile <4 x float> [[TMP0]], ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_pos (void)
 {
-// CHECK-LABEL: test_pos
-// CHECK: [[VAL:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: store volatile <4 x float> [[VAL]], ptr @ff
   ff = +ff2;
 }
 
+// CHECK-LABEL: define dso_local void @test_neg(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[FNEG:%.*]] = fneg <4 x float> [[TMP0]]
+// CHECK-NEXT:    store volatile <4 x float> [[FNEG]], ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_neg (void)
 {
-// CHECK-LABEL: test_neg
-// CHECK: [[VAL:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: %{{.*}} = fneg <4 x float> [[VAL]]
   ff = -ff2;
 }
 
+// CHECK-LABEL: define dso_local void @test_preinc(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[INC:%.*]] = fadd <4 x float> [[TMP0]], splat (float 1.000000e+00)
+// CHECK-NEXT:    store volatile <4 x float> [[INC]], ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_preinc (void)
 {
-// CHECK-LABEL: test_preinc
-// CHECK: [[VAL:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: %{{.*}} = fadd <4 x float> [[VAL]], splat (float 1.000000e+00)
   ++ff2;
 }
 
+// CHECK-LABEL: define dso_local void @test_postinc(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[INC:%.*]] = fadd <4 x float> [[TMP0]], splat (float 1.000000e+00)
+// CHECK-NEXT:    store volatile <4 x float> [[INC]], ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_postinc (void)
 {
-// CHECK-LABEL: test_postinc
-// CHECK: [[VAL:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: %{{.*}} = fadd <4 x float> [[VAL]], splat (float 1.000000e+00)
   ff2++;
 }
 
+// CHECK-LABEL: define dso_local void @test_predec(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[DEC:%.*]] = fadd <4 x float> [[TMP0]], splat (float -1.000000e+00)
+// CHECK-NEXT:    store volatile <4 x float> [[DEC]], ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_predec (void)
 {
-// CHECK-LABEL: test_predec
-// CHECK: [[VAL:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: %{{.*}} = fadd <4 x float> [[VAL]], splat (float -1.000000e+00)
   --ff2;
 }
 
+// CHECK-LABEL: define dso_local void @test_postdec(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[DEC:%.*]] = fadd <4 x float> [[TMP0]], splat (float -1.000000e+00)
+// CHECK-NEXT:    store volatile <4 x float> [[DEC]], ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_postdec (void)
 {
-// CHECK-LABEL: test_postdec
-// CHECK: [[VAL:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: %{{.*}} = fadd <4 x float> [[VAL]], splat (float -1.000000e+00)
   ff2--;
 }
 
+// CHECK-LABEL: define dso_local void @test_add(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[ADD:%.*]] = fadd <4 x float> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    store volatile <4 x float> [[ADD]], ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_add (void)
 {
-// CHECK-LABEL: test_add
-// CHECK: [[VAL1:%[^ ]+]] = load volatile <4 x float>, ptr @ff
-// CHECK: [[VAL2:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: %{{.*}} = fadd <4 x float> [[VAL1]], [[VAL2]]
   ff = ff + ff2;
 }
 
+// CHECK-LABEL: define dso_local void @test_add_assign(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[ADD:%.*]] = fadd <4 x float> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    store volatile <4 x float> [[ADD]], ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_add_assign (void)
 {
-// CHECK-LABEL: test_add_assign
-// CHECK: [[VAL2:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: [[VAL1:%[^ ]+]] = load volatile <4 x float>, ptr @ff
-// CHECK: %{{.*}} = fadd <4 x float> [[VAL2]], [[VAL1]]
   ff += ff2;
 }
 
+// CHECK-LABEL: define dso_local void @test_sub(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[SUB:%.*]] = fsub <4 x float> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    store volatile <4 x float> [[SUB]], ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_sub (void)
 {
-// CHECK-LABEL: test_sub
-// CHECK: [[VAL1:%[^ ]+]] = load volatile <4 x float>, ptr @ff
-// CHECK: [[VAL2:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: %{{.*}} = fsub <4 x float> [[VAL1]], [[VAL2]]
   ff = ff - ff2;
 }
 
+// CHECK-LABEL: define dso_local void @test_sub_assign(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[SUB:%.*]] = fsub <4 x float> [[TMP1]], [[TMP0]]
+// CHECK-NEXT:    store volatile <4 x float> [[SUB]], ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_sub_assign (void)
 {
-// CHECK-LABEL: test_sub_assign
-// CHECK: [[VAL2:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: [[VAL1:%[^ ]+]] = load volatile <4 x float>, ptr @ff
-// CHECK: %{{.*}} = fsub <4 x float> [[VAL1]], [[VAL2]]
   ff -= ff2;
 }
 
+// CHECK-LABEL: define dso_local void @test_mul(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[MUL:%.*]] = fmul <4 x float> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    store volatile <4 x float> [[MUL]], ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_mul (void)
 {
-// CHECK-LABEL: test_mul
-// CHECK: [[VAL1:%[^ ]+]] = load volatile <4 x float>, ptr @ff
-// CHECK: [[VAL2:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: %{{.*}} = fmul <4 x float> [[VAL1]], [[VAL2]]
   ff = ff * ff2;
 }
 
+// CHECK-LABEL: define dso_local void @test_mul_assign(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[MUL:%.*]] = fmul <4 x float> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    store volatile <4 x float> [[MUL]], ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_mul_assign (void)
 {
-// CHECK-LABEL: test_mul_assign
-// CHECK: [[VAL2:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: [[VAL1:%[^ ]+]] = load volatile <4 x float>, ptr @ff
-// CHECK: %{{.*}} = fmul <4 x float> [[VAL2]], [[VAL1]]
   ff *= ff2;
 }
 
+// CHECK-LABEL: define dso_local void @test_div(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[DIV:%.*]] = fdiv <4 x float> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    store volatile <4 x float> [[DIV]], ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_div (void)
 {
-// CHECK-LABEL: test_div
-// CHECK: [[VAL1:%[^ ]+]] = load volatile <4 x float>, ptr @ff
-// CHECK: [[VAL2:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: %{{.*}} = fdiv <4 x float> [[VAL1]], [[VAL2]]
   ff = ff / ff2;
 }
 
+// CHECK-LABEL: define dso_local void @test_div_assign(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[DIV:%.*]] = fdiv <4 x float> [[TMP1]], [[TMP0]]
+// CHECK-NEXT:    store volatile <4 x float> [[DIV]], ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_div_assign (void)
 {
-// CHECK-LABEL: test_div_assign
-// CHECK: [[VAL2:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: [[VAL1:%[^ ]+]] = load volatile <4 x float>, ptr @ff
-// CHECK: %{{.*}} = fdiv <4 x float> [[VAL1]], [[VAL2]]
   ff /= ff2;
 }
 
+// CHECK-LABEL: define dso_local void @test_cmpeq(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[CMP:%.*]] = fcmp oeq <4 x float> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT]], ptr @bi, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_cmpeq (void)
 {
-// CHECK-LABEL: test_cmpeq
-// CHECK: [[VAL1:%[^ ]+]] = load volatile <4 x float>, ptr @ff
-// CHECK: [[VAL2:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: [[CMP:%[^ ]+]] = fcmp oeq <4 x float> [[VAL1]], [[VAL2]]
-// CHECK: %{{.*}} = sext <4 x i1> [[CMP]] to <4 x i32>
   bi = ff == ff2;
 }
 
+// CHECK-LABEL: define dso_local void @test_cmpne(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[CMP:%.*]] = fcmp une <4 x float> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT]], ptr @bi, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_cmpne (void)
 {
-// CHECK-LABEL: test_cmpne
-// CHECK: [[VAL1:%[^ ]+]] = load volatile <4 x float>, ptr @ff
-// CHECK: [[VAL2:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: [[CMP:%[^ ]+]] = fcmp une <4 x float> [[VAL1]], [[VAL2]]
-// CHECK: %{{.*}} = sext <4 x i1> [[CMP]] to <4 x i32>
   bi = ff != ff2;
 }
 
+// CHECK-LABEL: define dso_local void @test_cmpge(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[CMP:%.*]] = fcmp oge <4 x float> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT]], ptr @bi, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_cmpge (void)
 {
-// CHECK-LABEL: test_cmpge
-// CHECK: [[VAL1:%[^ ]+]] = load volatile <4 x float>, ptr @ff
-// CHECK: [[VAL2:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: [[CMP:%[^ ]+]] = fcmp oge <4 x float> [[VAL1]], [[VAL2]]
-// CHECK: %{{.*}} = sext <4 x i1> [[CMP]] to <4 x i32>
   bi = ff >= ff2;
 }
 
+// CHECK-LABEL: define dso_local void @test_cmpgt(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[CMP:%.*]] = fcmp ogt <4 x float> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT]], ptr @bi, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_cmpgt (void)
 {
-// CHECK-LABEL: test_cmpgt
-// CHECK: [[VAL1:%[^ ]+]] = load volatile <4 x float>, ptr @ff
-// CHECK: [[VAL2:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: [[CMP:%[^ ]+]] = fcmp ogt <4 x float> [[VAL1]], [[VAL2]]
-// CHECK: %{{.*}} = sext <4 x i1> [[CMP]] to <4 x i32>
   bi = ff > ff2;
 }
 
+// CHECK-LABEL: define dso_local void @test_cmple(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[CMP:%.*]] = fcmp ole <4 x float> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT]], ptr @bi, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_cmple (void)
 {
-// CHECK-LABEL: test_cmple
-// CHECK: [[VAL1:%[^ ]+]] = load volatile <4 x float>, ptr @ff
-// CHECK: [[VAL2:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: [[CMP:%[^ ]+]] = fcmp ole <4 x float> [[VAL1]], [[VAL2]]
-// CHECK: %{{.*}} = sext <4 x i1> [[CMP]] to <4 x i32>
   bi = ff <= ff2;
 }
 
+// CHECK-LABEL: define dso_local void @test_cmplt(
+// CHECK-SAME: ) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load volatile <4 x float>, ptr @ff, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load volatile <4 x float>, ptr @ff2, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    [[CMP:%.*]] = fcmp olt <4 x float> [[TMP0]], [[TMP1]]
+// CHECK-NEXT:    [[SEXT:%.*]] = sext <4 x i1> [[CMP]] to <4 x i32>
+// CHECK-NEXT:    store volatile <4 x i32> [[SEXT]], ptr @bi, align 8, !tbaa [[TBAA3]]
+// CHECK-NEXT:    ret void
+//
 void test_cmplt (void)
 {
-// CHECK-LABEL: test_cmplt
-// CHECK: [[VAL1:%[^ ]+]] = load volatile <4 x float>, ptr @ff
-// CHECK: [[VAL2:%[^ ]+]] = load volatile <4 x float>, ptr @ff2
-// CHECK: [[CMP:%[^ ]+]] = fcmp olt <4 x float> [[VAL1]], [[VAL2]]
-// CHECK: %{{.*}} = sext <4 x i1> [[CMP]] to <4 x i32>
   bi = ff < ff2;
 }
 
+//.
+// CHECK: [[TBAA3]] = !{[[META4:![0-9]+]], [[META4]], i64 0}
+// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0}
+// CHECK: [[META5]] = !{!"Simple C/C++ TBAA"}
+//.


        


More information about the cfe-commits mailing list