[clang] [Clang][OpenMP] Support multidimensional target launch codegen with dims modifier (PR #239860)
Kevin Sala Penades via cfe-commits
cfe-commits at lists.llvm.org
Fri Nov 6 16:51:08 PST 2026
https://github.com/kevinsala created https://github.com/llvm/llvm-project/pull/239860
Propagate the dimensions of `num_teams`, `thread_limit`, and `num_threads` into target launch arguments. Apply thread limits and conditional parallel counts per dimension while preserving unidimensional behavior.
>From 8f6a3a9188ad70c46f084b4e416679142e15e699 Mon Sep 17 00:00:00 2001
From: Kevin Sala <salapenades1 at llnl.gov>
Date: Tue, 14 Jul 2026 11:34:48 -0700
Subject: [PATCH 1/2] [Clang][OpenMP] Update OpenMP CodeGen to support dims
modifier
---
clang/lib/CodeGen/CGOpenMPRuntime.cpp | 406 +++++++++++++++-----------
clang/lib/CodeGen/CGOpenMPRuntime.h | 26 +-
2 files changed, 254 insertions(+), 178 deletions(-)
diff --git a/clang/lib/CodeGen/CGOpenMPRuntime.cpp b/clang/lib/CodeGen/CGOpenMPRuntime.cpp
index d3a6ffee3ea748..efb9a7e6825be3 100644
--- a/clang/lib/CodeGen/CGOpenMPRuntime.cpp
+++ b/clang/lib/CodeGen/CGOpenMPRuntime.cpp
@@ -6558,13 +6558,10 @@ void CGOpenMPRuntime::computeMinAndMaxThreadsAndTeams(
llvm::OpenMPIRBuilder::TargetKernelDefaultAttrs &Attrs) {
assert(Attrs.MaxTeams.size() == 1 && Attrs.MaxThreads.size() == 1 &&
"invalid default attrs structure");
- int32_t &MaxTeamsVal = Attrs.MaxTeams.front();
- int32_t &MaxThreadsVal = Attrs.MaxThreads.front();
- getNumTeamsExprForTargetDirective(CGF, D, Attrs.MinTeams.front(),
- MaxTeamsVal);
- getNumThreadsExprForTargetDirective(CGF, D, MaxThreadsVal,
- /*UpperBoundOnly=*/true);
+ getNumTeamsExprForTargetDirective(CGF, D, Attrs.MinTeams, Attrs.MaxTeams);
+ getNumThreadsExprForTargetDirective(CGF, D, Attrs.MaxThreads,
+ /*UppersBoundOnly=*/true);
for (auto *C : D.getClausesOfKind<OMPXAttributeClause>()) {
for (auto *A : C->getAttrs()) {
@@ -6580,17 +6577,24 @@ void CGOpenMPRuntime::computeMinAndMaxThreadsAndTeams(
else
continue;
- Attrs.MinThreads.front() =
- std::max(Attrs.MinThreads.front(), AttrMinThreadsVal);
- if (AttrMaxThreadsVal > 0)
- MaxThreadsVal = MaxThreadsVal > 0
- ? std::min(MaxThreadsVal, AttrMaxThreadsVal)
- : AttrMaxThreadsVal;
- Attrs.MinTeams.front() =
- std::max(Attrs.MinTeams.front(), AttrMinBlocksVal);
- if (AttrMaxBlocksVal > 0)
- MaxTeamsVal = MaxTeamsVal > 0 ? std::min(MaxTeamsVal, AttrMaxBlocksVal)
- : AttrMaxBlocksVal;
+ // Ignore attributes if requesting multidimensional teams.
+ if (Attrs.MinThreads.size() == 1 && Attrs.MaxThreads.size() == 1) {
+ Attrs.MinThreads[0] = std::max(Attrs.MinThreads[0], AttrMinThreadsVal);
+ if (AttrMaxThreadsVal > 0)
+ Attrs.MaxThreads[0] =
+ Attrs.MaxThreads[0] > 0
+ ? std::min(Attrs.MaxThreads[0], AttrMaxThreadsVal)
+ : AttrMaxThreadsVal;
+ }
+ // Ignore attributes if requesting multidimensional leagues.
+ if (Attrs.MinTeams.size() == 1 && Attrs.MaxTeams.size() == 1) {
+ Attrs.MinTeams[0] = std::max(Attrs.MinTeams[0], AttrMinBlocksVal);
+ if (AttrMaxBlocksVal > 0)
+ Attrs.MaxTeams[0] =
+ Attrs.MaxTeams[0] > 0
+ ? std::min(Attrs.MaxTeams[0], AttrMaxBlocksVal)
+ : AttrMaxBlocksVal;
+ }
}
}
}
@@ -6689,9 +6693,14 @@ const Stmt *CGOpenMPRuntime::getSingleCompoundChild(ASTContext &Ctx,
return Child;
}
-const Expr *CGOpenMPRuntime::getNumTeamsExprForTargetDirective(
- CodeGenFunction &CGF, const OMPExecutableDirective &D, int32_t &MinTeamsVal,
- int32_t &MaxTeamsVal) {
+llvm::SmallVector<const Expr *>
+CGOpenMPRuntime::getNumTeamsExprForTargetDirective(
+ CodeGenFunction &CGF, const OMPExecutableDirective &D,
+ llvm::SmallVectorImpl<int32_t> &MinTeamsValues,
+ llvm::SmallVectorImpl<int32_t> &MaxTeamsValues) {
+ llvm::SmallVector<const Expr *> Exprs;
+ MinTeamsValues.assign(1, -1);
+ MaxTeamsValues.assign(1, -1);
OpenMPDirectiveKind DirectiveKind = D.getDirectiveKind();
assert(isOpenMPTargetExecutionDirective(DirectiveKind) &&
@@ -6707,24 +6716,29 @@ const Expr *CGOpenMPRuntime::getNumTeamsExprForTargetDirective(
dyn_cast_or_null<OMPExecutableDirective>(ChildStmt)) {
if (isOpenMPTeamsDirective(NestedDir->getDirectiveKind())) {
if (NestedDir->hasClausesOfKind<OMPNumTeamsClause>()) {
- const Expr *NumTeams = NestedDir->getSingleClause<OMPNumTeamsClause>()
- ->getNumTeams()
- .front();
- if (NumTeams->isIntegerConstantExpr(CGF.getContext()))
- if (auto Constant =
- NumTeams->getIntegerConstantExpr(CGF.getContext()))
- MinTeamsVal = MaxTeamsVal = Constant->getExtValue();
- return NumTeams;
+ auto NumTeamsArray =
+ NestedDir->getSingleClause<OMPNumTeamsClause>()->getNumTeams();
+ MinTeamsValues.resize(NumTeamsArray.size(), -1);
+ MaxTeamsValues.resize(NumTeamsArray.size(), -1);
+ for (const auto [Idx, NumTeams] : llvm::enumerate(NumTeamsArray)) {
+ if (NumTeams->isIntegerConstantExpr(CGF.getContext()))
+ if (auto Constant =
+ NumTeams->getIntegerConstantExpr(CGF.getContext()))
+ MinTeamsValues[Idx] = MaxTeamsValues[Idx] =
+ Constant->getExtValue();
+ Exprs.push_back(NumTeams);
+ }
+ return Exprs;
}
- MinTeamsVal = MaxTeamsVal = 0;
- return nullptr;
+ MinTeamsValues[0] = MaxTeamsValues[0] = 0;
+ return Exprs;
}
- MinTeamsVal = MaxTeamsVal = 1;
- return nullptr;
+ MinTeamsValues[0] = MaxTeamsValues[0] = 1;
+ return Exprs;
}
// A value of -1 is used to check if we need to emit no teams region
- MinTeamsVal = MaxTeamsVal = -1;
- return nullptr;
+ MinTeamsValues[0] = MaxTeamsValues[0] = -1;
+ return Exprs;
}
case OMPD_target_teams_loop:
case OMPD_target_teams:
@@ -6733,23 +6747,29 @@ const Expr *CGOpenMPRuntime::getNumTeamsExprForTargetDirective(
case OMPD_target_teams_distribute_parallel_for:
case OMPD_target_teams_distribute_parallel_for_simd: {
if (D.hasClausesOfKind<OMPNumTeamsClause>()) {
- const Expr *NumTeams =
- D.getSingleClause<OMPNumTeamsClause>()->getNumTeams().front();
- if (NumTeams->isIntegerConstantExpr(CGF.getContext()))
- if (auto Constant = NumTeams->getIntegerConstantExpr(CGF.getContext()))
- MinTeamsVal = MaxTeamsVal = Constant->getExtValue();
- return NumTeams;
- }
- MinTeamsVal = MaxTeamsVal = 0;
- return nullptr;
+ auto NumTeamsArray =
+ D.getSingleClause<OMPNumTeamsClause>()->getNumTeams();
+ MinTeamsValues.resize(NumTeamsArray.size(), -1);
+ MaxTeamsValues.resize(NumTeamsArray.size(), -1);
+ for (const auto [Idx, NumTeams] : llvm::enumerate(NumTeamsArray)) {
+ if (NumTeams->isIntegerConstantExpr(CGF.getContext()))
+ if (auto Constant =
+ NumTeams->getIntegerConstantExpr(CGF.getContext()))
+ MinTeamsValues[Idx] = MaxTeamsValues[Idx] = Constant->getExtValue();
+ Exprs.push_back(NumTeams);
+ }
+ return Exprs;
+ }
+ MinTeamsValues[0] = MaxTeamsValues[0] = 0;
+ return Exprs;
}
case OMPD_target_parallel:
case OMPD_target_parallel_for:
case OMPD_target_parallel_for_simd:
case OMPD_target_parallel_loop:
case OMPD_target_simd:
- MinTeamsVal = MaxTeamsVal = 1;
- return nullptr;
+ MinTeamsValues[0] = MaxTeamsValues[0] = 1;
+ return Exprs;
case OMPD_parallel:
case OMPD_for:
case OMPD_parallel_for:
@@ -6818,16 +6838,18 @@ const Expr *CGOpenMPRuntime::getNumTeamsExprForTargetDirective(
llvm_unreachable("Unexpected directive kind.");
}
-llvm::Value *CGOpenMPRuntime::emitNumTeamsForTargetDirective(
- CodeGenFunction &CGF, const OMPExecutableDirective &D) {
+void CGOpenMPRuntime::emitNumTeamsForTargetDirective(
+ CodeGenFunction &CGF, const OMPExecutableDirective &D,
+ llvm::SmallVectorImpl<llvm::Value *> &NumTeamsValues) {
assert(!CGF.getLangOpts().OpenMPIsTargetDevice &&
"Clauses associated with the teams directive expected to be emitted "
"only for the host!");
+ NumTeamsValues.clear();
+
CGBuilderTy &Bld = CGF.Builder;
- int32_t MinNT = -1, MaxNT = -1;
- const Expr *NumTeams =
- getNumTeamsExprForTargetDirective(CGF, D, MinNT, MaxNT);
- if (NumTeams != nullptr) {
+ llvm::SmallVector<int32_t> MinNT, MaxNT;
+ auto Exprs = getNumTeamsExprForTargetDirective(CGF, D, MinNT, MaxNT);
+ if (!Exprs.empty()) {
OpenMPDirectiveKind DirectiveKind = D.getDirectiveKind();
switch (DirectiveKind) {
@@ -6835,10 +6857,12 @@ llvm::Value *CGOpenMPRuntime::emitNumTeamsForTargetDirective(
const auto *CS = D.getInnermostCapturedStmt();
CGOpenMPInnerExprInfo CGInfo(CGF, *CS);
CodeGenFunction::CGCapturedStmtRAII CapInfoRAII(CGF, &CGInfo);
- llvm::Value *NumTeamsVal = CGF.EmitScalarExpr(NumTeams,
- /*IgnoreResultAssign*/ true);
- return Bld.CreateIntCast(NumTeamsVal, CGF.Int32Ty,
- /*isSigned=*/true);
+ for (const Expr *E : Exprs) {
+ llvm::Value *Val = CGF.EmitScalarExpr(E, /*IgnoreResultAssign*/ true);
+ NumTeamsValues.push_back(
+ Bld.CreateIntCast(Val, CGF.Int32Ty, /*isSigned=*/true));
+ }
+ return;
}
case OMPD_target_teams:
case OMPD_target_teams_distribute:
@@ -6846,10 +6870,12 @@ llvm::Value *CGOpenMPRuntime::emitNumTeamsForTargetDirective(
case OMPD_target_teams_distribute_parallel_for:
case OMPD_target_teams_distribute_parallel_for_simd: {
CodeGenFunction::RunCleanupsScope NumTeamsScope(CGF);
- llvm::Value *NumTeamsVal = CGF.EmitScalarExpr(NumTeams,
- /*IgnoreResultAssign*/ true);
- return Bld.CreateIntCast(NumTeamsVal, CGF.Int32Ty,
- /*isSigned=*/true);
+ for (const Expr *E : Exprs) {
+ llvm::Value *Val = CGF.EmitScalarExpr(E, /*IgnoreResultAssign*/ true);
+ NumTeamsValues.push_back(
+ Bld.CreateIntCast(Val, CGF.Int32Ty, /*isSigned=*/true));
+ }
+ return;
}
default:
break;
@@ -6857,7 +6883,9 @@ llvm::Value *CGOpenMPRuntime::emitNumTeamsForTargetDirective(
}
assert(MinNT == MaxNT && "Num threads ranges require handling here.");
- return llvm::ConstantInt::getSigned(CGF.Int32Ty, MinNT);
+
+ for (auto NT : MinNT)
+ NumTeamsValues.push_back(llvm::ConstantInt::getSigned(CGF.Int32Ty, NT));
}
/// Merge the thread count upper bound \p Val into \p UpperBound.
@@ -6874,13 +6902,50 @@ static void mergeThreadCountUpperBound(int32_t &UpperBound, int32_t Val) {
UpperBound = UpperBound > 0 ? std::min(UpperBound, Val) : Val;
}
+void getThreadExprsAndMergeUpperBound(
+ CodeGenFunction &CGF, llvm::ArrayRef<const Expr *> Exprs,
+ llvm::SmallVectorImpl<int32_t> &UpperBounds,
+ llvm::SmallVectorImpl<const Expr *> *ExprsPtr = nullptr) {
+ assert(!Exprs.empty() && "Invalid array of expressions.");
+ // All expressions must be constant to have an upperbound.
+ llvm::SmallVector<int32_t> Constants;
+ for (const Expr *E : Exprs)
+ if (E->isIntegerConstantExpr(CGF.getContext()))
+ if (auto Constant = E->getIntegerConstantExpr(CGF.getContext()))
+ Constants.push_back(static_cast<int32_t>(Constant->getZExtValue()));
+
+ if (Exprs.size() == Constants.size()) {
+ // All expressions are constant.
+ if (UpperBounds.empty()) {
+ // No bounds limiting clause were found before.
+ UpperBounds = Constants;
+ } else {
+ // Bounds were set previously. Truncate dimensions to the minimum and set
+ // the bounds to the minimum value in each dimension.
+ size_t MinSize = std::min(Exprs.size(), UpperBounds.size());
+ UpperBounds.truncate(MinSize);
+ for (size_t I = 0; I < MinSize; ++I)
+ mergeThreadCountUpperBound(UpperBounds[I], Constants[I]);
+ }
+ } else {
+ // Some of the expressions are not constant.
+ if (UpperBounds.empty())
+ UpperBounds.resize(Exprs.size(), 0);
+ else
+ UpperBounds.truncate(Exprs.size());
+ }
+ if (ExprsPtr)
+ (*ExprsPtr).assign(Exprs.begin(), Exprs.end());
+}
+
/// Check for a num threads constant value (stored in \p DefaultVal), or
/// expression (stored in \p E). If the value is conditional (via an if-clause),
/// store the condition in \p CondVal. If \p E, and \p CondVal respectively, are
/// nullptr, no expression evaluation is perfomed.
static void getNumThreads(CodeGenFunction &CGF, const CapturedStmt *CS,
- const Expr **E, int32_t &UpperBound,
- bool UpperBoundOnly, llvm::Value **CondVal) {
+ llvm::SmallVectorImpl<const Expr *> &Exprs,
+ llvm::SmallVectorImpl<int32_t> &UpperBounds,
+ bool UpperBoundsOnly, llvm::Value **CondVal) {
const Stmt *Child = CGOpenMPRuntime::getSingleCompoundChild(
CGF.getContext(), CS->getCapturedStmt());
const auto *Dir = dyn_cast_or_null<OMPExecutableDirective>(Child);
@@ -6906,7 +6971,7 @@ static void getNumThreads(CodeGenFunction &CGF, const CapturedStmt *CS,
bool Result;
if (CondExpr->EvaluateAsBooleanCondition(Result, CGF.getContext())) {
if (!Result) {
- UpperBound = 1;
+ UpperBounds.assign(1, 1);
return;
}
} else {
@@ -6934,18 +6999,12 @@ static void getNumThreads(CodeGenFunction &CGF, const CapturedStmt *CS,
CodeGenFunction::CGCapturedStmtRAII CapInfoRAII(CGF, &CGInfo);
const auto *NumThreadsClause =
Dir->getSingleClause<OMPNumThreadsClause>();
- const Expr *NTExpr = NumThreadsClause->getNumThreads().front();
- if (NTExpr->isIntegerConstantExpr(CGF.getContext()))
- if (auto Constant = NTExpr->getIntegerConstantExpr(CGF.getContext()))
- mergeThreadCountUpperBound(
- UpperBound, static_cast<int32_t>(Constant->getZExtValue()));
- // If we haven't found a upper bound, remember we saw a thread limiting
- // clause.
- if (UpperBound == -1)
- UpperBound = 0;
- if (!E)
+ llvm::ArrayRef<const Expr *> NTExprs = NumThreadsClause->getNumThreads();
+ getThreadExprsAndMergeUpperBound(CGF, NTExprs, UpperBounds);
+ if (UpperBoundsOnly)
return;
- CodeGenFunction::LexicalScope Scope(CGF, NTExpr->getSourceRange());
+
+ CodeGenFunction::LexicalScope Scope(CGF, NTExprs[0]->getSourceRange());
if (const auto *PreInit =
cast_or_null<DeclStmt>(NumThreadsClause->getPreInitStmt())) {
for (const auto *I : PreInit->decls()) {
@@ -6958,50 +7017,41 @@ static void getNumThreads(CodeGenFunction &CGF, const CapturedStmt *CS,
}
}
}
- *E = NTExpr;
+ Exprs.assign(NTExprs.begin(), NTExprs.end());
}
return;
}
if (isOpenMPSimdDirective(Dir->getDirectiveKind()))
- UpperBound = 1;
+ UpperBounds.assign(1, 1);
}
-const Expr *CGOpenMPRuntime::getNumThreadsExprForTargetDirective(
- CodeGenFunction &CGF, const OMPExecutableDirective &D, int32_t &UpperBound,
- bool UpperBoundOnly, llvm::Value **CondVal, const Expr **ThreadLimitExpr) {
- assert((!CGF.getLangOpts().OpenMPIsTargetDevice || UpperBoundOnly) &&
+llvm::SmallVector<const Expr *>
+CGOpenMPRuntime::getNumThreadsExprForTargetDirective(
+ CodeGenFunction &CGF, const OMPExecutableDirective &D,
+ llvm::SmallVectorImpl<int32_t> &UpperBounds, bool UpperBoundsOnly,
+ llvm::Value **CondVal,
+ llvm::SmallVectorImpl<const Expr *> *ThreadLimitExprs) {
+ assert((!CGF.getLangOpts().OpenMPIsTargetDevice || UpperBoundsOnly) &&
"Clauses associated with the teams directive expected to be emitted "
"only for the host!");
OpenMPDirectiveKind DirectiveKind = D.getDirectiveKind();
assert(isOpenMPTargetExecutionDirective(DirectiveKind) &&
"Expected target-based executable directive.");
- const Expr *NT = nullptr;
- const Expr **NTPtr = UpperBoundOnly ? nullptr : &NT;
+ llvm::SmallVector<const Expr *> NTExprs;
+ UpperBounds.clear();
- auto CheckForConstExpr = [&](const Expr *E, const Expr **EPtr) {
- if (E->isIntegerConstantExpr(CGF.getContext())) {
- if (auto Constant = E->getIntegerConstantExpr(CGF.getContext()))
- mergeThreadCountUpperBound(
- UpperBound, static_cast<int32_t>(Constant->getZExtValue()));
- }
- // If we haven't found a upper bound, remember we saw a thread limiting
- // clause.
- if (UpperBound == -1)
- UpperBound = 0;
- if (EPtr)
- *EPtr = E;
- };
-
- auto ReturnSequential = [&]() {
- UpperBound = 1;
- return NT;
+ auto SerializeBounds = [&]() { UpperBounds.assign(1, 1); };
+ auto NormalizeBounds = [&]() {
+ // Never return the upper bounds empty.
+ if (UpperBounds.empty())
+ UpperBounds.assign(1, -1);
};
switch (DirectiveKind) {
case OMPD_target: {
const CapturedStmt *CS = D.getInnermostCapturedStmt();
- getNumThreads(CGF, CS, NTPtr, UpperBound, UpperBoundOnly, CondVal);
+ getNumThreads(CGF, CS, NTExprs, UpperBounds, UpperBoundsOnly, CondVal);
const Stmt *Child = CGOpenMPRuntime::getSingleCompoundChild(
CGF.getContext(), CS->getCapturedStmt());
// TODO: The standard is not clear how to resolve two thread limit clauses,
@@ -7010,7 +7060,7 @@ const Expr *CGOpenMPRuntime::getNumThreadsExprForTargetDirective(
if (const auto *Dir = dyn_cast_or_null<OMPExecutableDirective>(Child)) {
if (const auto *TLC = Dir->getSingleClause<OMPThreadLimitClause>()) {
ThreadLimitClause = TLC;
- if (ThreadLimitExpr) {
+ if (ThreadLimitExprs) {
CGOpenMPInnerExprInfo CGInfo(CGF, *CS);
CodeGenFunction::CGCapturedStmtRAII CapInfoRAII(CGF, &CGInfo);
CodeGenFunction::LexicalScope Scope(
@@ -7032,8 +7082,8 @@ const Expr *CGOpenMPRuntime::getNumThreadsExprForTargetDirective(
}
}
if (ThreadLimitClause)
- CheckForConstExpr(ThreadLimitClause->getThreadLimit().front(),
- ThreadLimitExpr);
+ getThreadExprsAndMergeUpperBound(CGF, ThreadLimitClause->getThreadLimit(), UpperBounds,
+ ThreadLimitExprs);
if (const auto *Dir = dyn_cast_or_null<OMPExecutableDirective>(Child)) {
if (isOpenMPTeamsDirective(Dir->getDirectiveKind()) &&
!isOpenMPDistributeDirective(Dir->getDirectiveKind())) {
@@ -7044,10 +7094,10 @@ const Expr *CGOpenMPRuntime::getNumThreadsExprForTargetDirective(
// below does. Without this the upper bound of a construct written as
// 'target' / 'teams' / 'distribute parallel for' would stay at the
// default, while every combined spelling of the same construct honors
- // the clause. Only the bound is taken here: passing null for the
- // expression and the condition keeps this from emitting anything, so
- // the value the host passes to the kernel launch is left as it was.
- getNumThreads(CGF, CS, /*E=*/nullptr, UpperBound, UpperBoundOnly,
+ // the clause. Only the bounds are taken here expression and the
+ // condition keeps this from emitting anything, so the value the host
+ // passes to the kernel launch is left as it was.
+ getNumThreads(CGF, CS, NTExprs, UpperBounds, /*UpperBoundsOnly=*/true,
/*CondVal=*/nullptr);
const Stmt *Child = CGOpenMPRuntime::getSingleCompoundChild(
CGF.getContext(), CS->getCapturedStmt());
@@ -7055,41 +7105,46 @@ const Expr *CGOpenMPRuntime::getNumThreadsExprForTargetDirective(
}
if (Dir && isOpenMPParallelDirective(Dir->getDirectiveKind())) {
CS = Dir->getInnermostCapturedStmt();
- getNumThreads(CGF, CS, NTPtr, UpperBound, UpperBoundOnly, CondVal);
- } else if (Dir && isOpenMPSimdDirective(Dir->getDirectiveKind()))
- return ReturnSequential();
+ getNumThreads(CGF, CS, NTExprs, UpperBounds, UpperBoundsOnly, CondVal);
+ } else if (Dir && isOpenMPSimdDirective(Dir->getDirectiveKind())) {
+ SerializeBounds();
+ return NTExprs;
+ }
}
- return NT;
+ NormalizeBounds();
+ return NTExprs;
}
case OMPD_target_teams: {
if (D.hasClausesOfKind<OMPThreadLimitClause>()) {
CodeGenFunction::RunCleanupsScope ThreadLimitScope(CGF);
const auto *ThreadLimitClause = D.getSingleClause<OMPThreadLimitClause>();
- CheckForConstExpr(ThreadLimitClause->getThreadLimit().front(),
- ThreadLimitExpr);
+ getThreadExprsAndMergeUpperBound(CGF, ThreadLimitClause->getThreadLimit(), UpperBounds,
+ ThreadLimitExprs);
}
const CapturedStmt *CS = D.getInnermostCapturedStmt();
- getNumThreads(CGF, CS, NTPtr, UpperBound, UpperBoundOnly, CondVal);
+ getNumThreads(CGF, CS, NTExprs, UpperBounds, UpperBoundsOnly, CondVal);
const Stmt *Child = CGOpenMPRuntime::getSingleCompoundChild(
CGF.getContext(), CS->getCapturedStmt());
if (const auto *Dir = dyn_cast_or_null<OMPExecutableDirective>(Child)) {
if (Dir->getDirectiveKind() == OMPD_distribute) {
CS = Dir->getInnermostCapturedStmt();
- getNumThreads(CGF, CS, NTPtr, UpperBound, UpperBoundOnly, CondVal);
+ getNumThreads(CGF, CS, NTExprs, UpperBounds, UpperBoundsOnly, CondVal);
}
}
- return NT;
+ NormalizeBounds();
+ return NTExprs;
}
case OMPD_target_teams_distribute:
if (D.hasClausesOfKind<OMPThreadLimitClause>()) {
CodeGenFunction::RunCleanupsScope ThreadLimitScope(CGF);
const auto *ThreadLimitClause = D.getSingleClause<OMPThreadLimitClause>();
- CheckForConstExpr(ThreadLimitClause->getThreadLimit().front(),
- ThreadLimitExpr);
+ getThreadExprsAndMergeUpperBound(CGF, ThreadLimitClause->getThreadLimit(), UpperBounds,
+ ThreadLimitExprs);
}
- getNumThreads(CGF, D.getInnermostCapturedStmt(), NTPtr, UpperBound,
- UpperBoundOnly, CondVal);
- return NT;
+ getNumThreads(CGF, D.getInnermostCapturedStmt(), NTExprs, UpperBounds,
+ UpperBoundsOnly, CondVal);
+ NormalizeBounds();
+ return NTExprs;
case OMPD_target_teams_loop:
case OMPD_target_parallel_loop:
case OMPD_target_parallel:
@@ -7110,8 +7165,10 @@ const Expr *CGOpenMPRuntime::getNumThreadsExprForTargetDirective(
const Expr *Cond = IfClause->getCondition();
bool Result;
if (Cond->EvaluateAsBooleanCondition(Result, CGF.getContext())) {
- if (!Result)
- return ReturnSequential();
+ if (!Result) {
+ SerializeBounds();
+ return NTExprs;
+ }
} else {
CodeGenFunction::RunCleanupsScope Scope(CGF);
*CondVal = CGF.EvaluateExprAsBool(Cond);
@@ -7121,81 +7178,99 @@ const Expr *CGOpenMPRuntime::getNumThreadsExprForTargetDirective(
if (D.hasClausesOfKind<OMPThreadLimitClause>()) {
CodeGenFunction::RunCleanupsScope ThreadLimitScope(CGF);
const auto *ThreadLimitClause = D.getSingleClause<OMPThreadLimitClause>();
- CheckForConstExpr(ThreadLimitClause->getThreadLimit().front(),
- ThreadLimitExpr);
+ getThreadExprsAndMergeUpperBound(CGF, ThreadLimitClause->getThreadLimit(), UpperBounds,
+ ThreadLimitExprs);
}
if (D.hasClausesOfKind<OMPNumThreadsClause>()) {
CodeGenFunction::RunCleanupsScope NumThreadsScope(CGF);
const auto *NumThreadsClause = D.getSingleClause<OMPNumThreadsClause>();
- CheckForConstExpr(NumThreadsClause->getNumThreads().front(), nullptr);
- return NumThreadsClause->getNumThreads().front();
+ getThreadExprsAndMergeUpperBound(CGF, NumThreadsClause->getNumThreads(), UpperBounds);
+ NTExprs.assign(NumThreadsClause->getNumThreads().begin(),
+ NumThreadsClause->getNumThreads().end());
}
- return NT;
+ NormalizeBounds();
+ return NTExprs;
}
case OMPD_target_teams_distribute_simd:
case OMPD_target_simd:
- return ReturnSequential();
+ SerializeBounds();
+ return NTExprs;
default:
break;
}
llvm_unreachable("Unsupported directive kind.");
}
-llvm::Value *CGOpenMPRuntime::emitNumThreadsForTargetDirective(
- CodeGenFunction &CGF, const OMPExecutableDirective &D) {
- llvm::Value *NumThreadsVal = nullptr;
+void CGOpenMPRuntime::emitNumThreadsForTargetDirective(
+ CodeGenFunction &CGF, const OMPExecutableDirective &D,
+ llvm::SmallVectorImpl<llvm::Value *> &NumThreadsValues) {
llvm::Value *CondVal = nullptr;
- llvm::Value *ThreadLimitVal = nullptr;
- const Expr *ThreadLimitExpr = nullptr;
- int32_t UpperBound = -1;
+ llvm::SmallVector<int32_t> UpperBounds;
+ llvm::SmallVector<const Expr *> ThreadLimitExprs;
+ llvm::SmallVector<llvm::Value *> ThreadLimitValues;
- const Expr *NT = getNumThreadsExprForTargetDirective(
- CGF, D, UpperBound, /* UpperBoundOnly */ false, &CondVal,
- &ThreadLimitExpr);
+ auto NTExprs = getNumThreadsExprForTargetDirective(
+ CGF, D, UpperBounds, /* UpperBoundsOnly */ false, &CondVal,
+ &ThreadLimitExprs);
// Thread limit expressions are used below, emit them.
- if (ThreadLimitExpr) {
- ThreadLimitVal =
- CGF.EmitScalarExpr(ThreadLimitExpr, /*IgnoreResultAssign=*/true);
- ThreadLimitVal = CGF.Builder.CreateIntCast(ThreadLimitVal, CGF.Int32Ty,
- /*isSigned=*/false);
+ if (!ThreadLimitExprs.empty()) {
+ for (const Expr *TL : ThreadLimitExprs) {
+ llvm::Value *Val = CGF.EmitScalarExpr(TL, /*IgnoreResultAssign=*/true);
+ Val = CGF.Builder.CreateIntCast(Val, CGF.Int32Ty, /*isSigned=*/false);
+ ThreadLimitValues.push_back(Val);
+ }
}
- // Generate the num teams expression.
- if (UpperBound == 1) {
- NumThreadsVal = CGF.Builder.getInt32(UpperBound);
- } else if (NT) {
- NumThreadsVal = CGF.EmitScalarExpr(NT, /*IgnoreResultAssign=*/true);
- NumThreadsVal = CGF.Builder.CreateIntCast(NumThreadsVal, CGF.Int32Ty,
- /*isSigned=*/false);
- } else if (ThreadLimitVal) {
+ NumThreadsValues.clear();
+
+ // Generate the num threads expression.
+ if (!UpperBounds.empty() &&
+ llvm::all_of(UpperBounds, [](auto V) { return V == 1; })) {
+ for (int32_t UB : UpperBounds)
+ NumThreadsValues.push_back(CGF.Builder.getInt32(UB));
+ } else if (!NTExprs.empty()) {
+ for (const Expr *NT : NTExprs) {
+ llvm::Value *Val = CGF.EmitScalarExpr(NT, /*IgnoreResultAssign=*/true);
+ Val = CGF.Builder.CreateIntCast(Val, CGF.Int32Ty, /*isSigned=*/false);
+ NumThreadsValues.push_back(Val);
+ }
+ } else if (!ThreadLimitValues.empty()) {
// If we do not have a num threads value but a thread limit, replace the
// former with the latter. We know handled the thread limit expression.
- NumThreadsVal = ThreadLimitVal;
- ThreadLimitVal = nullptr;
+ NumThreadsValues = ThreadLimitValues;
+ ThreadLimitValues.clear();
} else {
// Default to "0" which means runtime choice.
- assert(!ThreadLimitVal && "Default not applicable with thread limit value");
- NumThreadsVal = CGF.Builder.getInt32(0);
+ assert(ThreadLimitValues.empty() &&
+ "Default not applicable with thread limit value");
+ NumThreadsValues.push_back(CGF.Builder.getInt32(0));
}
// Handle if clause. If if clause present, the number of threads is
// calculated as <cond> ? (<numthreads> ? <numthreads> : 0 ) : 1.
if (CondVal) {
CodeGenFunction::RunCleanupsScope Scope(CGF);
- NumThreadsVal = CGF.Builder.CreateSelect(CondVal, NumThreadsVal,
- CGF.Builder.getInt32(1));
+ for (llvm::Value *&NumThreadsVal : NumThreadsValues)
+ NumThreadsVal = CGF.Builder.CreateSelect(CondVal, NumThreadsVal,
+ CGF.Builder.getInt32(1));
}
// If the thread limit and num teams expression were present, take the
// minimum.
- if (ThreadLimitVal) {
- NumThreadsVal = CGF.Builder.CreateSelect(
- CGF.Builder.CreateICmpULT(ThreadLimitVal, NumThreadsVal),
- ThreadLimitVal, NumThreadsVal);
+ if (!ThreadLimitValues.empty()) {
+ assert(ThreadLimitValues.size() >= NumThreadsValues.size() &&
+ "Invalid num_threads clause.");
+ for (size_t Idx = 0; Idx < NumThreadsValues.size(); ++Idx) {
+ NumThreadsValues[Idx] = CGF.Builder.CreateSelect(
+ CGF.Builder.CreateICmpULT(ThreadLimitValues[Idx],
+ NumThreadsValues[Idx]),
+ ThreadLimitValues[Idx], NumThreadsValues[Idx]);
+ }
+ if (ThreadLimitValues.size() >= NumThreadsValues.size())
+ NumThreadsValues.resize(ThreadLimitValues.size(),
+ CGF.Builder.getInt32(1));
}
-
- return NumThreadsVal;
}
namespace {
@@ -11292,9 +11367,8 @@ static void emitTargetCallKernelLaunch(
emitClauseForBareTargetDirective<OMPThreadLimitClause>(CGF, D,
NumThreads);
} else {
- NumTeams.push_back(OMPRuntime->emitNumTeamsForTargetDirective(CGF, D));
- NumThreads.push_back(
- OMPRuntime->emitNumThreadsForTargetDirective(CGF, D));
+ OMPRuntime->emitNumTeamsForTargetDirective(CGF, D, NumTeams);
+ OMPRuntime->emitNumThreadsForTargetDirective(CGF, D, NumThreads);
}
llvm::Value *DeviceID = emitDeviceID(Device, CGF);
diff --git a/clang/lib/CodeGen/CGOpenMPRuntime.h b/clang/lib/CodeGen/CGOpenMPRuntime.h
index ae55295cadc14f..201eafc79330b4 100644
--- a/clang/lib/CodeGen/CGOpenMPRuntime.h
+++ b/clang/lib/CodeGen/CGOpenMPRuntime.h
@@ -628,28 +628,30 @@ class CGOpenMPRuntime {
/// have no associated teams construct.
///
/// Otherwise, return nullptr.
- const Expr *getNumTeamsExprForTargetDirective(CodeGenFunction &CGF,
- const OMPExecutableDirective &D,
- int32_t &MinTeamsVal,
- int32_t &MaxTeamsVal);
- llvm::Value *emitNumTeamsForTargetDirective(CodeGenFunction &CGF,
- const OMPExecutableDirective &D);
+ llvm::SmallVector<const Expr *> getNumTeamsExprForTargetDirective(
+ CodeGenFunction &CGF, const OMPExecutableDirective &D,
+ llvm::SmallVectorImpl<int32_t> &MinTeamsValues,
+ llvm::SmallVectorImpl<int32_t> &MaxTeamsValues);
+ void emitNumTeamsForTargetDirective(
+ CodeGenFunction &CGF, const OMPExecutableDirective &D,
+ llvm::SmallVectorImpl<llvm::Value *> &NumTeamsValues);
/// Check for a number of threads upper bound constant value (stored in \p
/// UpperBound), or expression (returned). If the value is conditional (via an
/// if-clause), store the condition in \p CondExpr. Similarly, a potential
/// thread limit expression is stored in \p ThreadLimitExpr. If \p
/// UpperBoundOnly is true, no expression evaluation is perfomed.
- const Expr *getNumThreadsExprForTargetDirective(
+ llvm::SmallVector<const Expr *> getNumThreadsExprForTargetDirective(
CodeGenFunction &CGF, const OMPExecutableDirective &D,
- int32_t &UpperBound, bool UpperBoundOnly,
- llvm::Value **CondExpr = nullptr, const Expr **ThreadLimitExpr = nullptr);
+ llvm::SmallVectorImpl<int32_t> &UpperBounds, bool UpperBoundsOnly,
+ llvm::Value **CondExpr = nullptr,
+ llvm::SmallVectorImpl<const Expr *> *ThreadLimitExprs = nullptr);
/// Emit an expression that denotes the number of threads a target region
/// shall use. Will generate "i32 0" to allow the runtime to choose.
- llvm::Value *
- emitNumThreadsForTargetDirective(CodeGenFunction &CGF,
- const OMPExecutableDirective &D);
+ void emitNumThreadsForTargetDirective(
+ CodeGenFunction &CGF, const OMPExecutableDirective &D,
+ llvm::SmallVectorImpl<llvm::Value *> &NumThreadsValues);
/// Return the trip count of loops associated with constructs / 'target teams
/// distribute' and 'teams distribute parallel for'. \param SizeEmitter Emits
>From 57177a899691494cb431063499808ae6cf94f034 Mon Sep 17 00:00:00 2001
From: Kevin Sala <salapenades1 at llnl.gov>
Date: Wed, 4 Nov 2026 17:01:41 -0800
Subject: [PATCH 2/2] Add codegen test
---
clang/test/OpenMP/dims_modifier_codegen.cpp | 14789 ++++++++++++++++++
1 file changed, 14789 insertions(+)
create mode 100644 clang/test/OpenMP/dims_modifier_codegen.cpp
diff --git a/clang/test/OpenMP/dims_modifier_codegen.cpp b/clang/test/OpenMP/dims_modifier_codegen.cpp
new file mode 100644
index 00000000000000..5b1ae5dfe6e579
--- /dev/null
+++ b/clang/test/OpenMP/dims_modifier_codegen.cpp
@@ -0,0 +1,14789 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _
+// Test host codegen.
+// RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=61 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1
+// RUN: %clang_cc1 -fopenmp -fopenmp-version=61 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -fopenmp-version=61 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1
+// RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=61 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3
+// RUN: %clang_cc1 -fopenmp -fopenmp-version=61 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -fopenmp-version=61 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK3
+
+// RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=61 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
+// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=61 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=61 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
+// RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=61 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
+// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=61 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=61 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
+
+// Test target codegen - host bc file has to be created first.
+// RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=61 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm-bc %s -o %t-ppc-host.bc
+// RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=61 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix=CHECK9
+// RUN: %clang_cc1 -fopenmp -fopenmp-version=61 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o %t %s
+// RUN: %clang_cc1 -fopenmp -fopenmp-version=61 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK9
+// RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=61 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm-bc %s -o %t-x86-host.bc
+// RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=61 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --check-prefix=CHECK11
+// RUN: %clang_cc1 -fopenmp -fopenmp-version=61 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-x86-host.bc -o %t %s
+// RUN: %clang_cc1 -fopenmp -fopenmp-version=61 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-x86-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK11
+
+// RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=61 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm-bc %s -o %t-ppc-host.bc
+// RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=61 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
+// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=61 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o %t %s
+// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=61 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
+// RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=61 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm-bc %s -o %t-x86-host.bc
+// RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=61 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
+// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=61 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-x86-host.bc -o %t %s
+// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=61 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-x86-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
+
+// expected-no-diagnostics
+
+#ifndef HEADER
+#define HEADER
+
+// Exercise the dimension counts passed to target kernel launches. The first
+// count is also passed directly to __tgt_target_kernel; the remaining counts
+// are available through the three-element arrays in the kernel arguments.
+
+// A single dimension with dims must retain the ordinary scalar behavior.
+void one_dimensional_modifiers() {
+#pragma omp target teams num_teams(dims(1): 3) thread_limit(dims(1): 16)
+ {}
+}
+
+void target_without_clauses() {
+#pragma omp target
+ {}
+}
+
+// Target teams can specify independent numbers of teams and threads.
+void target_teams_num_teams_only() {
+#pragma omp target teams num_teams(dims(3): 2, 3, 4)
+ {}
+}
+
+void target_teams_num_teams_dynamic(int teams_x, int teams_y) {
+#pragma omp target teams num_teams(dims(2): teams_x, teams_y)
+ {}
+}
+
+void target_teams_thread_limit_only() {
+#pragma omp target teams thread_limit(dims(3): 32, 8, 4)
+ {}
+}
+
+void target_teams_limits() {
+#pragma omp target teams num_teams(dims(2): 3, 2) thread_limit(dims(2): 16, 8)
+ {}
+}
+
+void target_thread_limit_only() {
+#pragma omp target thread_limit(dims(2): 32, 8)
+ {}
+}
+
+void target_nested_parallel(int threads_x, int threads_y) {
+#pragma omp target thread_limit(dims(2): 32, 8)
+#pragma omp parallel num_threads(dims(2): threads_x, threads_y)
+ {}
+}
+
+void target_nested_teams_parallel() {
+#pragma omp target
+#pragma omp teams thread_limit(dims(2): 32, 8)
+#pragma omp parallel num_threads(dims(2): 16, 4)
+ {}
+}
+
+void nested_teams_limit_overrides_target_limit() {
+#pragma omp target thread_limit(dims(2): 64, 16)
+#pragma omp teams thread_limit(dims(2): 32, 8)
+ {}
+}
+
+void target_teams_distribute() {
+#pragma omp target teams distribute thread_limit(dims(2): 32, 8)
+ for (int i = 0; i < 4; ++i) {}
+}
+
+void target_teams_distribute_parallel_for() {
+#pragma omp target teams distribute parallel for thread_limit(dims(2): 32, 8) num_threads(dims(2): 16, 4)
+ for (int i = 0; i < 4; ++i) {}
+}
+
+void target_teams_distribute_parallel_for_simd() {
+#pragma omp target teams distribute parallel for simd thread_limit(dims(2): 32, 8) num_threads(dims(2): 16, 4)
+ for (int i = 0; i < 4; ++i) {}
+}
+
+void target_teams_distribute_simd() {
+#pragma omp target teams distribute simd num_teams(dims(2): 3, 2) thread_limit(dims(2): 32, 8)
+ for (int i = 0; i < 4; ++i) {}
+}
+
+void target_teams_loop() {
+#pragma omp target teams loop num_teams(dims(2): 3, 2) thread_limit(dims(2): 32, 8)
+ for (int i = 0; i < 4; ++i) {}
+}
+
+// Combined parallel directives pass num_threads values to the launch.
+void target_parallel_thread_limit_only() {
+#pragma omp target parallel thread_limit(dims(2): 32, 8)
+ {}
+}
+
+void target_parallel_threads_dynamic(int threads_x, int threads_y) {
+#pragma omp target parallel num_threads(dims(2): threads_x, threads_y)
+ {}
+}
+
+void target_parallel_one_thread_dimension() {
+#pragma omp target parallel num_threads(dims(1): 16)
+ {}
+}
+
+void target_parallel_threads() {
+#pragma omp target parallel num_threads(dims(3): 8, 4, 2)
+ {}
+}
+
+void target_parallel_for() {
+#pragma omp target parallel for num_threads(dims(2): 16, 4)
+ for (int i = 0; i < 4; ++i) {}
+}
+
+void target_parallel_for_simd() {
+#pragma omp target parallel for simd num_threads(dims(2): 16, 4)
+ for (int i = 0; i < 4; ++i) {}
+}
+
+void target_parallel_loop() {
+#pragma omp target parallel loop num_threads(dims(2): 16, 4)
+ for (int i = 0; i < 4; ++i) {}
+}
+
+// Both clauses constrain each dimension separately.
+void target_parallel_dynamic(int limit_x, int limit_y, int threads_x,
+ int threads_y) {
+#pragma omp target parallel thread_limit(dims(2): limit_x, limit_y) num_threads(dims(2): threads_x, threads_y)
+ {}
+}
+
+void target_parallel_constant_limit(int threads_x, int threads_y) {
+#pragma omp target parallel thread_limit(dims(2): 32, 8) num_threads(dims(2): threads_x, threads_y)
+ {}
+}
+
+void target_parallel_dynamic_limit(int limit_x, int limit_y) {
+#pragma omp target parallel thread_limit(dims(2): limit_x, limit_y) num_threads(dims(2): 16, 4)
+ {}
+}
+
+void target_parallel_minimum() {
+#pragma omp target parallel thread_limit(dims(2): 32, 8) num_threads(dims(2): 16, 16)
+ {}
+}
+
+void target_parallel_three_dimensions() {
+#pragma omp target parallel thread_limit(dims(3): 32, 8, 4) num_threads(dims(3): 16, 16, 2)
+ {}
+}
+
+void target_parallel_all_ones() {
+#pragma omp target parallel num_threads(dims(2): 1, 1)
+ {}
+}
+
+void nested_target_teams_parallel() {
+#pragma omp target teams thread_limit(dims(2): 32, 8)
+#pragma omp parallel num_threads(dims(2): 16, 4)
+ {}
+}
+
+void nested_target_teams_distribute_parallel_for() {
+#pragma omp target teams thread_limit(dims(2): 32, 8)
+#pragma omp distribute parallel for num_threads(dims(2): 16, 4)
+ for (int i = 0; i < 4; ++i) {}
+}
+
+// A dynamic condition selects one thread in each dimension when false.
+void conditional_target_parallel(int cond, int threads_x, int threads_y) {
+#pragma omp target parallel if(parallel: cond) num_threads(dims(2): threads_x, threads_y)
+ {}
+}
+
+void conditional_target_parallel_with_limit(int cond) {
+#pragma omp target parallel if(parallel: cond) thread_limit(dims(2): 32, 8) num_threads(dims(2): 16, 4)
+ {}
+}
+
+void disabled_target_parallel() {
+#pragma omp target parallel if(parallel: 0) num_threads(dims(2): 16, 4)
+ {}
+}
+
+void enabled_target_parallel() {
+#pragma omp target parallel if(parallel: 1) num_threads(dims(2): 16, 4)
+ {}
+}
+
+// A longer thread_limit leaves the extra launch dimensions at one thread.
+
+void extra_limit_dimensions_1() {
+#pragma omp target parallel thread_limit(dims(2): 32, 8) num_threads(16)
+ {}
+}
+
+void extra_limit_dimensions_2() {
+#pragma omp target parallel thread_limit(dims(2): 32, 8) num_threads(dims(1): 16)
+ {}
+}
+
+void extra_limit_dimensions_3() {
+#pragma omp target parallel thread_limit(dims(3): 32, 8, 4) num_threads(16)
+ {}
+}
+
+void extra_limit_dimensions_4() {
+#pragma omp target parallel thread_limit(dims(3): 32, 8, 4) num_threads(dims(1): 16)
+ {}
+}
+
+void extra_limit_dimensions_5() {
+#pragma omp target parallel thread_limit(dims(3): 32, 8, 4) num_threads(dims(2): 16, 4)
+ {}
+}
+
+#endif
+// CHECK1-LABEL: define {{[^@]+}}@_Z25one_dimensional_modifiersv
+// CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 3, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 16, i32 0, i32 0], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1:[0-9]+]], i64 -1, i32 3, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z25one_dimensional_modifiersv_l47.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z25one_dimensional_modifiersv_l47(ptr null) #[[ATTR3:[0-9]+]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z25one_dimensional_modifiersv_l47
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR1:[0-9]+]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 3, i32 16)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z25one_dimensional_modifiersv_l47.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z25one_dimensional_modifiersv_l47.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2:[0-9]+]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z22target_without_clausesv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.1, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_without_clausesv_l52.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_without_clausesv_l52(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_without_clausesv_l52
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR2]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z27target_teams_num_teams_onlyv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.3, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 2, i32 3, i32 4], ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 2, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27target_teams_num_teams_onlyv_l58.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27target_teams_num_teams_onlyv_l58(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27target_teams_num_teams_onlyv_l58
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR4:[0-9]+]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 2, i32 0)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27target_teams_num_teams_onlyv_l58.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27target_teams_num_teams_onlyv_l58.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z30target_teams_num_teams_dynamicii
+// CHECK1-SAME: (i32 noundef signext [[TEAMS_X:%.*]], i32 noundef signext [[TEAMS_Y:%.*]]) #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[TEAMS_X_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[TEAMS_Y_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED2:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: store i32 [[TEAMS_X]], ptr [[TEAMS_X_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TEAMS_Y]], ptr [[TEAMS_Y_ADDR]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = load i32, ptr [[TEAMS_X_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TEAMS_Y_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR__CASTED2]], align 4
+// CHECK1-NEXT: [[TMP5:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED2]], align 8
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store i64 [[TMP3]], ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store i64 [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
+// CHECK1-NEXT: store i64 [[TMP5]], ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
+// CHECK1-NEXT: store i64 [[TMP5]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
+// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: [[TMP19:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP17]], 0
+// CHECK1-NEXT: [[TMP20:%.*]] = insertvalue [3 x i32] [[TMP19]], i32 [[TMP18]], 1
+// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP21]], align 4
+// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 3, ptr [[TMP22]], align 4
+// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP23]], align 8
+// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP24]], align 8
+// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.5, ptr [[TMP25]], align 8
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP26]], align 8
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP27]], align 8
+// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP28]], align 8
+// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP29]], align 8
+// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP30]], align 8
+// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [[TMP20]], ptr [[TMP31]], align 4
+// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP32]], align 4
+// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP33]], align 4
+// CHECK1-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 [[TMP17]], i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_num_teams_dynamicii_l63.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
+// CHECK1-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_num_teams_dynamicii_l63(i64 [[TMP3]], i64 [[TMP5]], ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_num_teams_dynamicii_l63
+// CHECK1-SAME: (i64 noundef [[DOTCAPTURE_EXPR_:%.*]], i64 noundef [[DOTCAPTURE_EXPR_1:%.*]], ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR2]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
+// CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_1]], ptr [[DOTCAPTURE_EXPR__ADDR2]], align 8
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
+// CHECK1-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP1]], i32 0)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_num_teams_dynamicii_l63.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_num_teams_dynamicii_l63.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z30target_teams_thread_limit_onlyv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.7, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 32, i32 8, i32 4], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 32, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_thread_limit_onlyv_l68.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_thread_limit_onlyv_l68(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_thread_limit_onlyv_l68
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR5:[0-9]+]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_thread_limit_onlyv_l68.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_thread_limit_onlyv_l68.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z19target_teams_limitsv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.9, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.10, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 3, i32 2, i32 0], ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 16, i32 8, i32 0], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 3, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_teams_limitsv_l73.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_teams_limitsv_l73(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_teams_limitsv_l73
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR1]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 3, i32 16)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_teams_limitsv_l73.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_teams_limitsv_l73.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z24target_thread_limit_onlyv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 1
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i64 64, i64 1, ptr @.omp_task_entry.)
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP6]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP9]], ptr align 8 [[TMP4]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T]], ptr [[TMP8]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP10]], ptr align 8 [[TMP5]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T]], ptr [[TMP8]], i32 0, i32 2
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP11]], ptr align 8 @.offload_sizes.11, i64 8, i1 false)
+// CHECK1-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry.(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK1-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_thread_limit_onlyv_l78
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR5]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map.
+// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6:[0-9]+]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T]], ptr [[TMP4]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry.
+// CHECK1-SAME: (i32 noundef signext [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7:[0-9]+]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP3]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META83:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META86:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META88:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META90:![0-9]+]])
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META92:![0-9]+]]
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: store ptr @.omp_task_privates_map., ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META93:![0-9]+]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META92]]
+// CHECK1-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK1-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.12, ptr [[TMP20]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP21]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP22]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP25]], align 4, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 32, i32 8, i32 0], ptr [[TMP26]], align 4, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META92]]
+// CHECK1-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 32, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_thread_limit_onlyv_l78.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK1-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK1-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__EXIT:%.*]]
+// CHECK1: omp_offload.failed.i:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_thread_limit_onlyv_l78(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[DOTOMP_OUTLINED__EXIT]]
+// CHECK1: .omp_outlined..exit:
+// CHECK1-NEXT: ret i32 0
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z22target_nested_parallelii
+// CHECK1-SAME: (i32 noundef signext [[THREADS_X:%.*]], i32 noundef signext [[THREADS_Y:%.*]]) #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[THREADS_X_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[THREADS_Y_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[THREADS_X_CASTED:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[THREADS_Y_CASTED:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8
+// CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_0:%.*]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store i32 [[THREADS_X]], ptr [[THREADS_X_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[THREADS_Y]], ptr [[THREADS_Y_ADDR]], align 4
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[THREADS_X_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP1]], ptr [[THREADS_X_CASTED]], align 4
+// CHECK1-NEXT: [[TMP2:%.*]] = load i64, ptr [[THREADS_X_CASTED]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[THREADS_Y_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP3]], ptr [[THREADS_Y_CASTED]], align 4
+// CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[THREADS_Y_CASTED]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store i64 [[TMP2]], ptr [[TMP5]], align 8
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store i64 [[TMP2]], ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
+// CHECK1-NEXT: store i64 [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
+// CHECK1-NEXT: store i64 [[TMP4]], ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
+// CHECK1-NEXT: store ptr null, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
+// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_0]], ptr [[AGG_CAPTURED]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[THREADS_X_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP17]], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_0]], ptr [[AGG_CAPTURED]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[THREADS_Y_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP19]], ptr [[TMP18]], align 4
+// CHECK1-NEXT: [[TMP20:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i64 120, i64 8, ptr @.omp_task_entry..17)
+// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2:%.*]], ptr [[TMP20]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP21]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP23:%.*]] = load ptr, ptr [[TMP22]], align 8
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP23]], ptr align 4 [[AGG_CAPTURED]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2]], ptr [[TMP20]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1:%.*]], ptr [[TMP24]], i32 0, i32 0
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP25]], ptr align 8 [[TMP14]], i64 24, i1 false)
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1]], ptr [[TMP24]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP26]], ptr align 8 [[TMP15]], i64 24, i1 false)
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1]], ptr [[TMP24]], i32 0, i32 2
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP27]], ptr align 8 @.offload_sizes.13, i64 24, i1 false)
+// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1]], ptr [[TMP24]], i32 0, i32 3
+// CHECK1-NEXT: [[TMP29:%.*]] = load i32, ptr [[THREADS_X_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP29]], ptr [[TMP28]], align 8
+// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1]], ptr [[TMP24]], i32 0, i32 4
+// CHECK1-NEXT: [[TMP31:%.*]] = load i32, ptr [[THREADS_Y_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP31]], ptr [[TMP30]], align 4
+// CHECK1-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP20]])
+// CHECK1-NEXT: [[TMP32:%.*]] = call i32 @.omp_task_entry..17(i32 [[TMP0]], ptr [[TMP20]]) #[[ATTR3]]
+// CHECK1-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP20]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_nested_parallelii_l83
+// CHECK1-SAME: (i64 noundef [[THREADS_X:%.*]], i64 noundef [[THREADS_Y:%.*]], ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR9:[0-9]+]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[THREADS_X_ADDR:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[THREADS_Y_ADDR:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store i64 [[THREADS_X]], ptr [[THREADS_X_ADDR]], align 8
+// CHECK1-NEXT: store i64 [[THREADS_Y]], ptr [[THREADS_Y_ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[THREADS_X_ADDR]], align 4
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP1]])
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_nested_parallelii_l83.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_nested_parallelii_l83.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10:[0-9]+]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map..16
+// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]], ptr noalias noundef [[TMP4:%.*]], ptr noalias noundef [[TMP5:%.*]]) #[[ATTR6]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR4:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR5:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[DOTADDR4]], align 8
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTADDR5]], align 8
+// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1]], ptr [[TMP6]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR4]], align 8
+// CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1]], ptr [[TMP6]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTADDR5]], align 8
+// CHECK1-NEXT: store ptr [[TMP11]], ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1]], ptr [[TMP6]], i32 0, i32 3
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1]], ptr [[TMP6]], i32 0, i32 4
+// CHECK1-NEXT: [[TMP16:%.*]] = load ptr, ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP16]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry..17
+// CHECK1-SAME: (i32 noundef signext [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR3_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR4_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[THREADS_X_CASTED_I:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[THREADS_Y_CASTED_I:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2]], ptr [[TMP3]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META94:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META97:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META99:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META101:![0-9]+]])
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META103:![0-9]+]]
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: store ptr @.omp_task_privates_map..16, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR3_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR4_I]]) #[[ATTR3]], !inline_history [[META104:![0-9]+]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP15:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR3_I]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP16:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR4_I]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META103]]
+// CHECK1-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP17]], i32 32)
+// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP12]], align 4
+// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP13]], align 4
+// CHECK1-NEXT: [[TMP20:%.*]] = icmp ult i32 32, [[TMP18]]
+// CHECK1-NEXT: [[TMP21:%.*]] = select i1 [[TMP20]], i32 32, i32 [[TMP18]]
+// CHECK1-NEXT: [[TMP22:%.*]] = icmp ult i32 8, [[TMP19]]
+// CHECK1-NEXT: [[TMP23:%.*]] = select i1 [[TMP22]], i32 8, i32 [[TMP19]]
+// CHECK1-NEXT: [[TMP24:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP21]], 0
+// CHECK1-NEXT: [[TMP25:%.*]] = insertvalue [3 x i32] [[TMP24]], i32 [[TMP23]], 1
+// CHECK1-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK1-NEXT: store i32 3, ptr [[TMP26]], align 4, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP27]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP28]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP29]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.14, ptr [[TMP30]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP31]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP32]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP33]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP34]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP35]], align 4, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [[TMP25]], ptr [[TMP36]], align 4, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP37]], align 4, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 [[TMP21]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_nested_parallelii_l83.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK1-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0
+// CHECK1-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__15_EXIT:%.*]]
+// CHECK1: omp_offload.failed.i:
+// CHECK1-NEXT: [[TMP40:%.*]] = load i32, ptr [[TMP12]], align 4
+// CHECK1-NEXT: store i32 [[TMP40]], ptr [[THREADS_X_CASTED_I]], align 4, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP41:%.*]] = load i64, ptr [[THREADS_X_CASTED_I]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP42:%.*]] = load i32, ptr [[TMP13]], align 4
+// CHECK1-NEXT: store i32 [[TMP42]], ptr [[THREADS_Y_CASTED_I]], align 4, !noalias [[META103]]
+// CHECK1-NEXT: [[TMP43:%.*]] = load i64, ptr [[THREADS_Y_CASTED_I]], align 8, !noalias [[META103]]
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_nested_parallelii_l83(i64 [[TMP41]], i64 [[TMP43]], ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[DOTOMP_OUTLINED__15_EXIT]]
+// CHECK1: .omp_outlined..15.exit:
+// CHECK1-NEXT: ret i32 0
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z28target_nested_teams_parallelv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.18, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.19, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 32, i32 8, i32 0], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 32, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR11:[0-9]+]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89.omp_outlined.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89.omp_outlined.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z41nested_teams_limit_overrides_target_limitv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_3:%.*]], align 1
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i64 64, i64 1, ptr @.omp_task_entry..24)
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_5:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_5]], ptr [[TMP6]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_4:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP9]], ptr align 8 [[TMP4]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_4]], ptr [[TMP8]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP10]], ptr align 8 [[TMP5]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_4]], ptr [[TMP8]], i32 0, i32 2
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP11]], ptr align 8 @.offload_sizes.20, i64 8, i1 false)
+// CHECK1-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry..24(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK1-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41nested_teams_limit_overrides_target_limitv_l96
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR5]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41nested_teams_limit_overrides_target_limitv_l96.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41nested_teams_limit_overrides_target_limitv_l96.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map..23
+// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_4:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_4]], ptr [[TMP4]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_4]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry..24
+// CHECK1-SAME: (i32 noundef signext [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_5:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_5]], ptr [[TMP3]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META105:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META108:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META110:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META112:![0-9]+]])
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META114:![0-9]+]]
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: store ptr @.omp_task_privates_map..23, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META115:![0-9]+]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META114]]
+// CHECK1-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 64)
+// CHECK1-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.21, ptr [[TMP20]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP21]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP22]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP25]], align 4, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 32, i32 8, i32 0], ptr [[TMP26]], align 4, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META114]]
+// CHECK1-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 32, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41nested_teams_limit_overrides_target_limitv_l96.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK1-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK1-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__22_EXIT:%.*]]
+// CHECK1: omp_offload.failed.i:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41nested_teams_limit_overrides_target_limitv_l96(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[DOTOMP_OUTLINED__22_EXIT]]
+// CHECK1: .omp_outlined..22.exit:
+// CHECK1-NEXT: ret i32 0
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z23target_teams_distributev
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.25, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.26, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 4, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 32, i32 8, i32 0], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 32, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_teams_distributev_l102.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_teams_distributev_l102(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_teams_distributev_l102
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR5]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_teams_distributev_l102.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_teams_distributev_l102.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK1: cond.true:
+// CHECK1-NEXT: br label [[COND_END:%.*]]
+// CHECK1: cond.false:
+// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: br label [[COND_END]]
+// CHECK1: cond.end:
+// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK1: omp.inner.for.cond:
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK1: omp.inner.for.body:
+// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK1: omp.body.continue:
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK1: omp.inner.for.inc:
+// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK1-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK1: omp.loop.exit:
+// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z36target_teams_distribute_parallel_forv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.27, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.28, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 4, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 16, i32 4, i32 0], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR11]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: store i32 3, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK1: cond.true:
+// CHECK1-NEXT: br label [[COND_END:%.*]]
+// CHECK1: cond.false:
+// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: br label [[COND_END]]
+// CHECK1: cond.end:
+// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK1: omp.inner.for.cond:
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK1: omp.inner.for.body:
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16)
+// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107.omp_outlined.omp_outlined, i64 [[TMP8]], i64 [[TMP10]])
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK1: omp.inner.for.inc:
+// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
+// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK1: omp.loop.exit:
+// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107.omp_outlined.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
+// CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
+// CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
+// CHECK1-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
+// CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
+// CHECK1-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 3
+// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK1: cond.true:
+// CHECK1-NEXT: br label [[COND_END:%.*]]
+// CHECK1: cond.false:
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: br label [[COND_END]]
+// CHECK1: cond.end:
+// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
+// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK1: omp.inner.for.cond:
+// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
+// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK1: omp.inner.for.body:
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
+// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK1: omp.body.continue:
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK1: omp.inner.for.inc:
+// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
+// CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK1: omp.loop.exit:
+// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z41target_teams_distribute_parallel_for_simdv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.29, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.30, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 4, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 16, i32 4, i32 0], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR11]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: store i32 3, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK1: cond.true:
+// CHECK1-NEXT: br label [[COND_END:%.*]]
+// CHECK1: cond.false:
+// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: br label [[COND_END]]
+// CHECK1: cond.end:
+// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK1: omp.inner.for.cond:
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP116:![0-9]+]]
+// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK1: omp.inner.for.body:
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16), !llvm.access.group [[ACC_GRP116]]
+// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112.omp_outlined.omp_outlined, i64 [[TMP8]], i64 [[TMP10]]), !llvm.access.group [[ACC_GRP116]]
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK1: omp.inner.for.inc:
+// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
+// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP117:![0-9]+]]
+// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK1: omp.loop.exit:
+// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
+// CHECK1-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK1: .omp.final.then:
+// CHECK1-NEXT: store i32 4, ptr [[I]], align 4
+// CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
+// CHECK1: .omp.final.done:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112.omp_outlined.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
+// CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
+// CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
+// CHECK1-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
+// CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
+// CHECK1-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 3
+// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK1: cond.true:
+// CHECK1-NEXT: br label [[COND_END:%.*]]
+// CHECK1: cond.false:
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: br label [[COND_END]]
+// CHECK1: cond.end:
+// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
+// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK1: omp.inner.for.cond:
+// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120:![0-9]+]]
+// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP120]]
+// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
+// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK1: omp.inner.for.body:
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120]]
+// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
+// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP120]]
+// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK1: omp.body.continue:
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK1: omp.inner.for.inc:
+// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120]]
+// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
+// CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120]]
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP121:![0-9]+]]
+// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK1: omp.loop.exit:
+// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
+// CHECK1-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK1: .omp.final.then:
+// CHECK1-NEXT: store i32 4, ptr [[I]], align 4
+// CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
+// CHECK1: .omp.final.done:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z28target_teams_distribute_simdv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.31, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.32, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 4, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 3, i32 2, i32 0], ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 3, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_teams_distribute_simdv_l117.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_teams_distribute_simdv_l117(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_teams_distribute_simdv_l117
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR12:[0-9]+]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 3, i32 32)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_teams_distribute_simdv_l117.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_teams_distribute_simdv_l117.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK1: cond.true:
+// CHECK1-NEXT: br label [[COND_END:%.*]]
+// CHECK1: cond.false:
+// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: br label [[COND_END]]
+// CHECK1: cond.end:
+// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK1: omp.inner.for.cond:
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123:![0-9]+]]
+// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK1: omp.inner.for.body:
+// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK1: omp.body.continue:
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK1: omp.inner.for.inc:
+// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK1-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP124:![0-9]+]]
+// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK1: omp.loop.exit:
+// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP10:%.*]] = icmp ne i32 [[TMP9]], 0
+// CHECK1-NEXT: br i1 [[TMP10]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK1: .omp.final.then:
+// CHECK1-NEXT: store i32 4, ptr [[I]], align 4
+// CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
+// CHECK1: .omp.final.done:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z17target_teams_loopv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.33, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.34, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 4, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 3, i32 2, i32 0], ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 32, i32 8, i32 0], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 3, i32 32, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR13:[0-9]+]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 3, i32 32)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: store i32 3, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK1: cond.true:
+// CHECK1-NEXT: br label [[COND_END:%.*]]
+// CHECK1: cond.false:
+// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: br label [[COND_END]]
+// CHECK1: cond.end:
+// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK1: omp.inner.for.cond:
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK1: omp.inner.for.body:
+// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122.omp_outlined.omp_outlined, i64 [[TMP8]], i64 [[TMP10]])
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK1: omp.inner.for.inc:
+// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
+// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK1: omp.loop.exit:
+// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122.omp_outlined.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
+// CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
+// CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
+// CHECK1-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
+// CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
+// CHECK1-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 3
+// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK1: cond.true:
+// CHECK1-NEXT: br label [[COND_END:%.*]]
+// CHECK1: cond.false:
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: br label [[COND_END]]
+// CHECK1: cond.end:
+// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
+// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK1: omp.inner.for.cond:
+// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
+// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK1: omp.inner.for.body:
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
+// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK1: omp.body.continue:
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK1: omp.inner.for.inc:
+// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
+// CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK1: omp.loop.exit:
+// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z33target_parallel_thread_limit_onlyv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_6:%.*]], align 1
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i64 64, i64 1, ptr @.omp_task_entry..39)
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_8:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_8]], ptr [[TMP6]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_7:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP9]], ptr align 8 [[TMP4]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_7]], ptr [[TMP8]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP10]], ptr align 8 [[TMP5]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_7]], ptr [[TMP8]], i32 0, i32 2
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP11]], ptr align 8 @.offload_sizes.35, i64 8, i1 false)
+// CHECK1-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry..39(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK1-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z33target_parallel_thread_limit_onlyv_l128
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR9]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z33target_parallel_thread_limit_onlyv_l128.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z33target_parallel_thread_limit_onlyv_l128.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map..38
+// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_7:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_7]], ptr [[TMP4]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_7]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry..39
+// CHECK1-SAME: (i32 noundef signext [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_8:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_8]], ptr [[TMP3]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META126:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META129:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META131:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META133:![0-9]+]])
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META135:![0-9]+]]
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: store ptr @.omp_task_privates_map..38, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META136:![0-9]+]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META135]]
+// CHECK1-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK1-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.36, ptr [[TMP20]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP21]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP22]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP25]], align 4, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 32, i32 8, i32 0], ptr [[TMP26]], align 4, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META135]]
+// CHECK1-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 32, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z33target_parallel_thread_limit_onlyv_l128.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK1-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK1-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__37_EXIT:%.*]]
+// CHECK1: omp_offload.failed.i:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z33target_parallel_thread_limit_onlyv_l128(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[DOTOMP_OUTLINED__37_EXIT]]
+// CHECK1: .omp_outlined..37.exit:
+// CHECK1-NEXT: ret i32 0
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z31target_parallel_threads_dynamicii
+// CHECK1-SAME: (i32 noundef signext [[THREADS_X:%.*]], i32 noundef signext [[THREADS_Y:%.*]]) #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[THREADS_X_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[THREADS_Y_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED2:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: store i32 [[THREADS_X]], ptr [[THREADS_X_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[THREADS_Y]], ptr [[THREADS_Y_ADDR]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = load i32, ptr [[THREADS_X_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[THREADS_Y_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR__CASTED2]], align 4
+// CHECK1-NEXT: [[TMP5:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED2]], align 8
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store i64 [[TMP3]], ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store i64 [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
+// CHECK1-NEXT: store i64 [[TMP5]], ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
+// CHECK1-NEXT: store i64 [[TMP5]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
+// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: [[TMP19:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP17]], 0
+// CHECK1-NEXT: [[TMP20:%.*]] = insertvalue [3 x i32] [[TMP19]], i32 [[TMP18]], 1
+// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP21]], align 4
+// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 3, ptr [[TMP22]], align 4
+// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP23]], align 8
+// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP24]], align 8
+// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.40, ptr [[TMP25]], align 8
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.41, ptr [[TMP26]], align 8
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP27]], align 8
+// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP28]], align 8
+// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP29]], align 8
+// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP30]], align 8
+// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP31]], align 4
+// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [[TMP20]], ptr [[TMP32]], align 4
+// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP33]], align 4
+// CHECK1-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 [[TMP17]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z31target_parallel_threads_dynamicii_l133.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
+// CHECK1-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z31target_parallel_threads_dynamicii_l133(i64 [[TMP3]], i64 [[TMP5]], ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z31target_parallel_threads_dynamicii_l133
+// CHECK1-SAME: (i64 noundef [[DOTCAPTURE_EXPR_:%.*]], i64 noundef [[DOTCAPTURE_EXPR_1:%.*]], ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR14:[0-9]+]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
+// CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_1]], ptr [[DOTCAPTURE_EXPR__ADDR2]], align 8
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP1]])
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z31target_parallel_threads_dynamicii_l133.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z31target_parallel_threads_dynamicii_l133.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z36target_parallel_one_thread_dimensionv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.42, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.43, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 16, i32 0, i32 0], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_parallel_one_thread_dimensionv_l138.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_parallel_one_thread_dimensionv_l138(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_parallel_one_thread_dimensionv_l138
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15:[0-9]+]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_parallel_one_thread_dimensionv_l138.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_parallel_one_thread_dimensionv_l138.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z23target_parallel_threadsv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.44, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.45, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 8, i32 4, i32 2], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 8, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_threadsv_l143.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_threadsv_l143(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_threadsv_l143
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR16:[0-9]+]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 8)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_threadsv_l143.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_threadsv_l143.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z19target_parallel_forv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.46, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.47, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 16, i32 4, i32 0], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_parallel_forv_l148.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_parallel_forv_l148(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_parallel_forv_l148
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_parallel_forv_l148.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_parallel_forv_l148.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK1: cond.true:
+// CHECK1-NEXT: br label [[COND_END:%.*]]
+// CHECK1: cond.false:
+// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: br label [[COND_END]]
+// CHECK1: cond.end:
+// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK1: omp.inner.for.cond:
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK1: omp.inner.for.body:
+// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK1: omp.body.continue:
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK1: omp.inner.for.inc:
+// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK1-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK1: omp.loop.exit:
+// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z24target_parallel_for_simdv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.48, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.49, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 16, i32 4, i32 0], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_for_simdv_l153.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_for_simdv_l153(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_for_simdv_l153
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_for_simdv_l153.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_for_simdv_l153.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK1: cond.true:
+// CHECK1-NEXT: br label [[COND_END:%.*]]
+// CHECK1: cond.false:
+// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: br label [[COND_END]]
+// CHECK1: cond.end:
+// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK1: omp.inner.for.cond:
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137:![0-9]+]]
+// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK1: omp.inner.for.body:
+// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK1: omp.body.continue:
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK1: omp.inner.for.inc:
+// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK1-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP138:![0-9]+]]
+// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK1: omp.loop.exit:
+// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP10:%.*]] = icmp ne i32 [[TMP9]], 0
+// CHECK1-NEXT: br i1 [[TMP10]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK1: .omp.final.then:
+// CHECK1-NEXT: store i32 4, ptr [[I]], align 4
+// CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
+// CHECK1: .omp.final.done:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z20target_parallel_loopv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.50, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.51, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 16, i32 4, i32 0], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20target_parallel_loopv_l158.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20target_parallel_loopv_l158(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20target_parallel_loopv_l158
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20target_parallel_loopv_l158.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20target_parallel_loopv_l158.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK1: cond.true:
+// CHECK1-NEXT: br label [[COND_END:%.*]]
+// CHECK1: cond.false:
+// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: br label [[COND_END]]
+// CHECK1: cond.end:
+// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK1: omp.inner.for.cond:
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK1: omp.inner.for.body:
+// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK1: omp.body.continue:
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK1: omp.inner.for.inc:
+// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK1-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK1: omp.loop.exit:
+// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z23target_parallel_dynamiciiii
+// CHECK1-SAME: (i32 noundef signext [[LIMIT_X:%.*]], i32 noundef signext [[LIMIT_Y:%.*]], i32 noundef signext [[THREADS_X:%.*]], i32 noundef signext [[THREADS_Y:%.*]]) #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[LIMIT_X_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[LIMIT_Y_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[THREADS_X_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[THREADS_Y_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED4:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED5:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED6:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x ptr], align 8
+// CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_9:%.*]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store i32 [[LIMIT_X]], ptr [[LIMIT_X_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[LIMIT_Y]], ptr [[LIMIT_Y_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[THREADS_X]], ptr [[THREADS_X_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[THREADS_Y]], ptr [[THREADS_Y_ADDR]], align 4
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[LIMIT_X_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[LIMIT_Y_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[THREADS_X_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
+// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[THREADS_Y_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_3]], align 4
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: store i32 [[TMP7]], ptr [[DOTCAPTURE_EXPR__CASTED4]], align 4
+// CHECK1-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED4]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
+// CHECK1-NEXT: store i32 [[TMP9]], ptr [[DOTCAPTURE_EXPR__CASTED5]], align 4
+// CHECK1-NEXT: [[TMP10:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED5]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3]], align 4
+// CHECK1-NEXT: store i32 [[TMP11]], ptr [[DOTCAPTURE_EXPR__CASTED6]], align 4
+// CHECK1-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED6]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store i64 [[TMP6]], ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store i64 [[TMP6]], ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP15]], align 8
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
+// CHECK1-NEXT: store i64 [[TMP8]], ptr [[TMP16]], align 8
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
+// CHECK1-NEXT: store i64 [[TMP8]], ptr [[TMP17]], align 8
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
+// CHECK1-NEXT: store ptr null, ptr [[TMP18]], align 8
+// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
+// CHECK1-NEXT: store i64 [[TMP10]], ptr [[TMP19]], align 8
+// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
+// CHECK1-NEXT: store i64 [[TMP10]], ptr [[TMP20]], align 8
+// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
+// CHECK1-NEXT: store ptr null, ptr [[TMP21]], align 8
+// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
+// CHECK1-NEXT: store i64 [[TMP12]], ptr [[TMP22]], align 8
+// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3
+// CHECK1-NEXT: store i64 [[TMP12]], ptr [[TMP23]], align 8
+// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
+// CHECK1-NEXT: store ptr null, ptr [[TMP24]], align 8
+// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr null, ptr [[TMP25]], align 8
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr null, ptr [[TMP26]], align 8
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
+// CHECK1-NEXT: store ptr null, ptr [[TMP27]], align 8
+// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_9]], ptr [[AGG_CAPTURED]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK1-NEXT: store i32 [[TMP31]], ptr [[TMP30]], align 4
+// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_9]], ptr [[AGG_CAPTURED]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: store i32 [[TMP33]], ptr [[TMP32]], align 4
+// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_9]], ptr [[AGG_CAPTURED]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
+// CHECK1-NEXT: store i32 [[TMP35]], ptr [[TMP34]], align 4
+// CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_9]], ptr [[AGG_CAPTURED]], i32 0, i32 3
+// CHECK1-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3]], align 4
+// CHECK1-NEXT: store i32 [[TMP37]], ptr [[TMP36]], align 4
+// CHECK1-NEXT: [[TMP38:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i64 160, i64 16, ptr @.omp_task_entry..56)
+// CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_11:%.*]], ptr [[TMP38]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP39]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP41:%.*]] = load ptr, ptr [[TMP40]], align 8
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP41]], ptr align 4 [[AGG_CAPTURED]], i64 16, i1 false)
+// CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_11]], ptr [[TMP38]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_10:%.*]], ptr [[TMP42]], i32 0, i32 0
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP43]], ptr align 8 [[TMP28]], i64 40, i1 false)
+// CHECK1-NEXT: [[TMP44:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_10]], ptr [[TMP42]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP44]], ptr align 8 [[TMP29]], i64 40, i1 false)
+// CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_10]], ptr [[TMP42]], i32 0, i32 2
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP45]], ptr align 8 @.offload_sizes.52, i64 40, i1 false)
+// CHECK1-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP38]])
+// CHECK1-NEXT: [[TMP46:%.*]] = call i32 @.omp_task_entry..56(i32 [[TMP0]], ptr [[TMP38]]) #[[ATTR3]]
+// CHECK1-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP38]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_dynamiciiii_l165
+// CHECK1-SAME: (i64 noundef [[DOTCAPTURE_EXPR_:%.*]], i64 noundef [[DOTCAPTURE_EXPR_1:%.*]], i64 noundef [[DOTCAPTURE_EXPR_3:%.*]], i64 noundef [[DOTCAPTURE_EXPR_5:%.*]], ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR14]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR4:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR6:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
+// CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_1]], ptr [[DOTCAPTURE_EXPR__ADDR2]], align 8
+// CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_3]], ptr [[DOTCAPTURE_EXPR__ADDR4]], align 8
+// CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_5]], ptr [[DOTCAPTURE_EXPR__ADDR6]], align 8
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR4]], align 4
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP1]])
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_dynamiciiii_l165.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_dynamiciiii_l165.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map..55
+// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_10:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_10]], ptr [[TMP4]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_10]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry..56
+// CHECK1-SAME: (i32 noundef signext [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED_I:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED3_I:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED4_I:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED5_I:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_11:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_11]], ptr [[TMP3]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META140:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META143:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META145:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META147:![0-9]+]])
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META149:![0-9]+]]
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: store ptr @.omp_task_privates_map..55, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META150:![0-9]+]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_9:%.*]], ptr [[TMP9]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_9]], ptr [[TMP9]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_9]], ptr [[TMP9]], i32 0, i32 3
+// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP9]], align 4
+// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META149]]
+// CHECK1-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP19]], i32 [[TMP18]])
+// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP9]], align 4
+// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP24:%.*]] = icmp ult i32 [[TMP20]], [[TMP22]]
+// CHECK1-NEXT: [[TMP25:%.*]] = select i1 [[TMP24]], i32 [[TMP20]], i32 [[TMP22]]
+// CHECK1-NEXT: [[TMP26:%.*]] = icmp ult i32 [[TMP21]], [[TMP23]]
+// CHECK1-NEXT: [[TMP27:%.*]] = select i1 [[TMP26]], i32 [[TMP21]], i32 [[TMP23]]
+// CHECK1-NEXT: [[TMP28:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP25]], 0
+// CHECK1-NEXT: [[TMP29:%.*]] = insertvalue [3 x i32] [[TMP28]], i32 [[TMP27]], 1
+// CHECK1-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK1-NEXT: store i32 5, ptr [[TMP30]], align 4, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP31]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP32]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP33]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.53, ptr [[TMP34]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP35]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP36]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP37]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP38]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP39]], align 4, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [[TMP29]], ptr [[TMP40]], align 4, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP41]], align 4, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 [[TMP25]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_dynamiciiii_l165.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK1-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
+// CHECK1-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__54_EXIT:%.*]]
+// CHECK1: omp_offload.failed.i:
+// CHECK1-NEXT: [[TMP44:%.*]] = load i32, ptr [[TMP9]], align 4
+// CHECK1-NEXT: store i32 [[TMP44]], ptr [[DOTCAPTURE_EXPR__CASTED_I]], align 4, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP45:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED_I]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP46:%.*]] = load i32, ptr [[TMP15]], align 4
+// CHECK1-NEXT: store i32 [[TMP46]], ptr [[DOTCAPTURE_EXPR__CASTED3_I]], align 4, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP47:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED3_I]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP48:%.*]] = load i32, ptr [[TMP16]], align 4
+// CHECK1-NEXT: store i32 [[TMP48]], ptr [[DOTCAPTURE_EXPR__CASTED4_I]], align 4, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP49:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED4_I]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP50:%.*]] = load i32, ptr [[TMP17]], align 4
+// CHECK1-NEXT: store i32 [[TMP50]], ptr [[DOTCAPTURE_EXPR__CASTED5_I]], align 4, !noalias [[META149]]
+// CHECK1-NEXT: [[TMP51:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED5_I]], align 8, !noalias [[META149]]
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_dynamiciiii_l165(i64 [[TMP45]], i64 [[TMP47]], i64 [[TMP49]], i64 [[TMP51]], ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[DOTOMP_OUTLINED__54_EXIT]]
+// CHECK1: .omp_outlined..54.exit:
+// CHECK1-NEXT: ret i32 0
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z30target_parallel_constant_limitii
+// CHECK1-SAME: (i32 noundef signext [[THREADS_X:%.*]], i32 noundef signext [[THREADS_Y:%.*]]) #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[THREADS_X_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[THREADS_Y_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED2:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8
+// CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_12:%.*]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store i32 [[THREADS_X]], ptr [[THREADS_X_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[THREADS_Y]], ptr [[THREADS_Y_ADDR]], align 4
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[THREADS_X_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[THREADS_Y_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK1-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTCAPTURE_EXPR__CASTED2]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED2]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store i64 [[TMP4]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store i64 [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
+// CHECK1-NEXT: store i64 [[TMP6]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
+// CHECK1-NEXT: store i64 [[TMP6]], ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
+// CHECK1-NEXT: store ptr null, ptr [[TMP15]], align 8
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_12]], ptr [[AGG_CAPTURED]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK1-NEXT: store i32 [[TMP19]], ptr [[TMP18]], align 4
+// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_12]], ptr [[AGG_CAPTURED]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: store i32 [[TMP21]], ptr [[TMP20]], align 4
+// CHECK1-NEXT: [[TMP22:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i64 112, i64 8, ptr @.omp_task_entry..61)
+// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_14:%.*]], ptr [[TMP22]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP23]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP25:%.*]] = load ptr, ptr [[TMP24]], align 8
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP25]], ptr align 4 [[AGG_CAPTURED]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_14]], ptr [[TMP22]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_13:%.*]], ptr [[TMP26]], i32 0, i32 0
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP27]], ptr align 8 [[TMP16]], i64 24, i1 false)
+// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_13]], ptr [[TMP26]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP28]], ptr align 8 [[TMP17]], i64 24, i1 false)
+// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_13]], ptr [[TMP26]], i32 0, i32 2
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP29]], ptr align 8 @.offload_sizes.57, i64 24, i1 false)
+// CHECK1-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP22]])
+// CHECK1-NEXT: [[TMP30:%.*]] = call i32 @.omp_task_entry..61(i32 [[TMP0]], ptr [[TMP22]]) #[[ATTR3]]
+// CHECK1-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP22]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_parallel_constant_limitii_l170
+// CHECK1-SAME: (i64 noundef [[DOTCAPTURE_EXPR_:%.*]], i64 noundef [[DOTCAPTURE_EXPR_1:%.*]], ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR9]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
+// CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_1]], ptr [[DOTCAPTURE_EXPR__ADDR2]], align 8
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP1]])
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_parallel_constant_limitii_l170.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_parallel_constant_limitii_l170.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map..60
+// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_13:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_13]], ptr [[TMP4]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_13]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry..61
+// CHECK1-SAME: (i32 noundef signext [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED_I:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED3_I:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_14:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_14]], ptr [[TMP3]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META151:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META154:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META156:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META158:![0-9]+]])
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META160:![0-9]+]]
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: store ptr @.omp_task_privates_map..60, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META161:![0-9]+]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_12:%.*]], ptr [[TMP9]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META160]]
+// CHECK1-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP16]], i32 32)
+// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[TMP9]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ult i32 32, [[TMP17]]
+// CHECK1-NEXT: [[TMP20:%.*]] = select i1 [[TMP19]], i32 32, i32 [[TMP17]]
+// CHECK1-NEXT: [[TMP21:%.*]] = icmp ult i32 8, [[TMP18]]
+// CHECK1-NEXT: [[TMP22:%.*]] = select i1 [[TMP21]], i32 8, i32 [[TMP18]]
+// CHECK1-NEXT: [[TMP23:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP20]], 0
+// CHECK1-NEXT: [[TMP24:%.*]] = insertvalue [3 x i32] [[TMP23]], i32 [[TMP22]], 1
+// CHECK1-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK1-NEXT: store i32 3, ptr [[TMP25]], align 4, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP26]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP27]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP28]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.58, ptr [[TMP29]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP30]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP31]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP32]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP33]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP34]], align 4, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [[TMP24]], ptr [[TMP35]], align 4, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP36]], align 4, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 [[TMP20]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_parallel_constant_limitii_l170.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK1-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
+// CHECK1-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__59_EXIT:%.*]]
+// CHECK1: omp_offload.failed.i:
+// CHECK1-NEXT: [[TMP39:%.*]] = load i32, ptr [[TMP9]], align 4
+// CHECK1-NEXT: store i32 [[TMP39]], ptr [[DOTCAPTURE_EXPR__CASTED_I]], align 4, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP40:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED_I]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP41:%.*]] = load i32, ptr [[TMP15]], align 4
+// CHECK1-NEXT: store i32 [[TMP41]], ptr [[DOTCAPTURE_EXPR__CASTED3_I]], align 4, !noalias [[META160]]
+// CHECK1-NEXT: [[TMP42:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED3_I]], align 8, !noalias [[META160]]
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_parallel_constant_limitii_l170(i64 [[TMP40]], i64 [[TMP42]], ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[DOTOMP_OUTLINED__59_EXIT]]
+// CHECK1: .omp_outlined..59.exit:
+// CHECK1-NEXT: ret i32 0
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z29target_parallel_dynamic_limitii
+// CHECK1-SAME: (i32 noundef signext [[LIMIT_X:%.*]], i32 noundef signext [[LIMIT_Y:%.*]]) #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[LIMIT_X_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[LIMIT_Y_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED2:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8
+// CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_15:%.*]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store i32 [[LIMIT_X]], ptr [[LIMIT_X_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[LIMIT_Y]], ptr [[LIMIT_Y_ADDR]], align 4
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[LIMIT_X_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[LIMIT_Y_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK1-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTCAPTURE_EXPR__CASTED2]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED2]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store i64 [[TMP4]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store i64 [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
+// CHECK1-NEXT: store i64 [[TMP6]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
+// CHECK1-NEXT: store i64 [[TMP6]], ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
+// CHECK1-NEXT: store ptr null, ptr [[TMP15]], align 8
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_15]], ptr [[AGG_CAPTURED]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK1-NEXT: store i32 [[TMP19]], ptr [[TMP18]], align 4
+// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_15]], ptr [[AGG_CAPTURED]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: store i32 [[TMP21]], ptr [[TMP20]], align 4
+// CHECK1-NEXT: [[TMP22:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i64 112, i64 8, ptr @.omp_task_entry..66)
+// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_17:%.*]], ptr [[TMP22]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP23]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP25:%.*]] = load ptr, ptr [[TMP24]], align 8
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP25]], ptr align 4 [[AGG_CAPTURED]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_17]], ptr [[TMP22]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_16:%.*]], ptr [[TMP26]], i32 0, i32 0
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP27]], ptr align 8 [[TMP16]], i64 24, i1 false)
+// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_16]], ptr [[TMP26]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP28]], ptr align 8 [[TMP17]], i64 24, i1 false)
+// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_16]], ptr [[TMP26]], i32 0, i32 2
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP29]], ptr align 8 @.offload_sizes.62, i64 24, i1 false)
+// CHECK1-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP22]])
+// CHECK1-NEXT: [[TMP30:%.*]] = call i32 @.omp_task_entry..66(i32 [[TMP0]], ptr [[TMP22]]) #[[ATTR3]]
+// CHECK1-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP22]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z29target_parallel_dynamic_limitii_l175
+// CHECK1-SAME: (i64 noundef [[DOTCAPTURE_EXPR_:%.*]], i64 noundef [[DOTCAPTURE_EXPR_1:%.*]], ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
+// CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_1]], ptr [[DOTCAPTURE_EXPR__ADDR2]], align 8
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z29target_parallel_dynamic_limitii_l175.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z29target_parallel_dynamic_limitii_l175.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map..65
+// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_16:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_16]], ptr [[TMP4]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_16]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry..66
+// CHECK1-SAME: (i32 noundef signext [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED_I:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED3_I:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_17:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_17]], ptr [[TMP3]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META162:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META165:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META167:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META169:![0-9]+]])
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META171:![0-9]+]]
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: store ptr @.omp_task_privates_map..65, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META172:![0-9]+]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_15:%.*]], ptr [[TMP9]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP9]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META171]]
+// CHECK1-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP17]], i32 [[TMP16]])
+// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP9]], align 4
+// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP20:%.*]] = icmp ult i32 [[TMP18]], 16
+// CHECK1-NEXT: [[TMP21:%.*]] = select i1 [[TMP20]], i32 [[TMP18]], i32 16
+// CHECK1-NEXT: [[TMP22:%.*]] = icmp ult i32 [[TMP19]], 4
+// CHECK1-NEXT: [[TMP23:%.*]] = select i1 [[TMP22]], i32 [[TMP19]], i32 4
+// CHECK1-NEXT: [[TMP24:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP21]], 0
+// CHECK1-NEXT: [[TMP25:%.*]] = insertvalue [3 x i32] [[TMP24]], i32 [[TMP23]], 1
+// CHECK1-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK1-NEXT: store i32 3, ptr [[TMP26]], align 4, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP27]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP28]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP29]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.63, ptr [[TMP30]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP31]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP32]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP33]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP34]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP35]], align 4, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [[TMP25]], ptr [[TMP36]], align 4, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP37]], align 4, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 [[TMP21]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z29target_parallel_dynamic_limitii_l175.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK1-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0
+// CHECK1-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__64_EXIT:%.*]]
+// CHECK1: omp_offload.failed.i:
+// CHECK1-NEXT: [[TMP40:%.*]] = load i32, ptr [[TMP9]], align 4
+// CHECK1-NEXT: store i32 [[TMP40]], ptr [[DOTCAPTURE_EXPR__CASTED_I]], align 4, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP41:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED_I]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP42:%.*]] = load i32, ptr [[TMP15]], align 4
+// CHECK1-NEXT: store i32 [[TMP42]], ptr [[DOTCAPTURE_EXPR__CASTED3_I]], align 4, !noalias [[META171]]
+// CHECK1-NEXT: [[TMP43:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED3_I]], align 8, !noalias [[META171]]
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z29target_parallel_dynamic_limitii_l175(i64 [[TMP41]], i64 [[TMP43]], ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[DOTOMP_OUTLINED__64_EXIT]]
+// CHECK1: .omp_outlined..64.exit:
+// CHECK1-NEXT: ret i32 0
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z23target_parallel_minimumv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_18:%.*]], align 1
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i64 64, i64 1, ptr @.omp_task_entry..71)
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_20:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_20]], ptr [[TMP6]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_19:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP9]], ptr align 8 [[TMP4]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_19]], ptr [[TMP8]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP10]], ptr align 8 [[TMP5]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_19]], ptr [[TMP8]], i32 0, i32 2
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP11]], ptr align 8 @.offload_sizes.67, i64 8, i1 false)
+// CHECK1-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry..71(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK1-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_minimumv_l180
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_minimumv_l180.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_minimumv_l180.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map..70
+// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_19:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_19]], ptr [[TMP4]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_19]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry..71
+// CHECK1-SAME: (i32 noundef signext [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_20:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_20]], ptr [[TMP3]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META173:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META176:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META178:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META180:![0-9]+]])
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META182:![0-9]+]]
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: store ptr @.omp_task_privates_map..70, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META183:![0-9]+]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META182]]
+// CHECK1-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK1-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.68, ptr [[TMP20]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP21]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP22]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP25]], align 4, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 16, i32 8, i32 0], ptr [[TMP26]], align 4, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META182]]
+// CHECK1-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_minimumv_l180.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK1-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK1-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__69_EXIT:%.*]]
+// CHECK1: omp_offload.failed.i:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_minimumv_l180(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[DOTOMP_OUTLINED__69_EXIT]]
+// CHECK1: .omp_outlined..69.exit:
+// CHECK1-NEXT: ret i32 0
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z32target_parallel_three_dimensionsv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_21:%.*]], align 1
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i64 64, i64 1, ptr @.omp_task_entry..76)
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_23:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_23]], ptr [[TMP6]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_22:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP9]], ptr align 8 [[TMP4]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_22]], ptr [[TMP8]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP10]], ptr align 8 [[TMP5]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_22]], ptr [[TMP8]], i32 0, i32 2
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP11]], ptr align 8 @.offload_sizes.72, i64 8, i1 false)
+// CHECK1-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry..76(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK1-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32target_parallel_three_dimensionsv_l185
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32target_parallel_three_dimensionsv_l185.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32target_parallel_three_dimensionsv_l185.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map..75
+// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_22:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_22]], ptr [[TMP4]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_22]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry..76
+// CHECK1-SAME: (i32 noundef signext [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_23:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_23]], ptr [[TMP3]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META184:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META187:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META189:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META191:![0-9]+]])
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META193:![0-9]+]]
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: store ptr @.omp_task_privates_map..75, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META194:![0-9]+]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META193]]
+// CHECK1-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK1-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.73, ptr [[TMP20]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP21]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP22]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP25]], align 4, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 16, i32 8, i32 2], ptr [[TMP26]], align 4, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META193]]
+// CHECK1-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32target_parallel_three_dimensionsv_l185.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK1-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK1-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__74_EXIT:%.*]]
+// CHECK1: omp_offload.failed.i:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32target_parallel_three_dimensionsv_l185(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[DOTOMP_OUTLINED__74_EXIT]]
+// CHECK1: .omp_outlined..74.exit:
+// CHECK1-NEXT: ret i32 0
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z24target_parallel_all_onesv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.77, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.78, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 1, i32 0], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_all_onesv_l190.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_all_onesv_l190(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_all_onesv_l190
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR17:[0-9]+]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 1)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_all_onesv_l190.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_all_onesv_l190.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z28nested_target_teams_parallelv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.79, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.80, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 16, i32 4, i32 0], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR11]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195.omp_outlined.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195.omp_outlined.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z43nested_target_teams_distribute_parallel_forv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.81, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.82, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 4, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 16, i32 4, i32 0], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR11]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: store i32 3, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK1: cond.true:
+// CHECK1-NEXT: br label [[COND_END:%.*]]
+// CHECK1: cond.false:
+// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: br label [[COND_END]]
+// CHECK1: cond.end:
+// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK1: omp.inner.for.cond:
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK1: omp.inner.for.body:
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16)
+// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201.omp_outlined.omp_outlined, i64 [[TMP8]], i64 [[TMP10]])
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK1: omp.inner.for.inc:
+// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
+// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK1: omp.loop.exit:
+// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201.omp_outlined.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
+// CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
+// CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
+// CHECK1-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
+// CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
+// CHECK1-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
+// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 3
+// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK1: cond.true:
+// CHECK1-NEXT: br label [[COND_END:%.*]]
+// CHECK1: cond.false:
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: br label [[COND_END]]
+// CHECK1: cond.end:
+// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
+// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK1-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK1: omp.inner.for.cond:
+// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
+// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK1: omp.inner.for.body:
+// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
+// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK1: omp.body.continue:
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK1: omp.inner.for.inc:
+// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
+// CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
+// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK1: omp.inner.for.end:
+// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK1: omp.loop.exit:
+// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z27conditional_target_paralleliii
+// CHECK1-SAME: (i32 noundef signext [[COND:%.*]], i32 noundef signext [[THREADS_X:%.*]], i32 noundef signext [[THREADS_Y:%.*]]) #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[COND_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[THREADS_X_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[THREADS_Y_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED4:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED5:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x ptr], align 8
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: store i32 [[COND]], ptr [[COND_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[THREADS_X]], ptr [[THREADS_X_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[THREADS_Y]], ptr [[THREADS_Y_ADDR]], align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = load i32, ptr [[COND_ADDR]], align 4
+// CHECK1-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
+// CHECK1-NEXT: [[STOREDV:%.*]] = zext i1 [[TOBOOL]] to i8
+// CHECK1-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[THREADS_X_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[THREADS_Y_ADDR]], align 4
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_2]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK1-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP3]], 0
+// CHECK1-NEXT: [[STOREDV3:%.*]] = zext i1 [[LOADEDV]] to i8
+// CHECK1-NEXT: store i8 [[STOREDV3]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1
+// CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTCAPTURE_EXPR__CASTED4]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED4]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
+// CHECK1-NEXT: store i32 [[TMP7]], ptr [[DOTCAPTURE_EXPR__CASTED5]], align 4
+// CHECK1-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED5]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store i64 [[TMP4]], ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store i64 [[TMP4]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
+// CHECK1-NEXT: store i64 [[TMP6]], ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
+// CHECK1-NEXT: store i64 [[TMP6]], ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
+// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
+// CHECK1-NEXT: store i64 [[TMP8]], ptr [[TMP15]], align 8
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
+// CHECK1-NEXT: store i64 [[TMP8]], ptr [[TMP16]], align 8
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
+// CHECK1-NEXT: store ptr null, ptr [[TMP17]], align 8
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr null, ptr [[TMP18]], align 8
+// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr null, ptr [[TMP19]], align 8
+// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
+// CHECK1-NEXT: store ptr null, ptr [[TMP20]], align 8
+// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP23:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK1-NEXT: [[LOADEDV6:%.*]] = icmp ne i8 [[TMP23]], 0
+// CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK1-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
+// CHECK1-NEXT: [[TMP26:%.*]] = select i1 [[LOADEDV6]], i32 [[TMP24]], i32 1
+// CHECK1-NEXT: [[TMP27:%.*]] = select i1 [[LOADEDV6]], i32 [[TMP25]], i32 1
+// CHECK1-NEXT: [[TMP28:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP26]], 0
+// CHECK1-NEXT: [[TMP29:%.*]] = insertvalue [3 x i32] [[TMP28]], i32 [[TMP27]], 1
+// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP30]], align 4
+// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 4, ptr [[TMP31]], align 4
+// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP21]], ptr [[TMP32]], align 8
+// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP22]], ptr [[TMP33]], align 8
+// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.83, ptr [[TMP34]], align 8
+// CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.84, ptr [[TMP35]], align 8
+// CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP36]], align 8
+// CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP37]], align 8
+// CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP38]], align 8
+// CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP39]], align 8
+// CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP40]], align 4
+// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [[TMP29]], ptr [[TMP41]], align 4
+// CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP42]], align 4
+// CHECK1-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 [[TMP26]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
+// CHECK1-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208(i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP8]], ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208
+// CHECK1-SAME: (i64 noundef [[DOTCAPTURE_EXPR_:%.*]], i64 noundef [[DOTCAPTURE_EXPR_1:%.*]], i64 noundef [[DOTCAPTURE_EXPR_3:%.*]], ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR14]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR4:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
+// CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_1]], ptr [[DOTCAPTURE_EXPR__ADDR2]], align 8
+// CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_3]], ptr [[DOTCAPTURE_EXPR__ADDR4]], align 8
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR2]], align 4
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP1]])
+// CHECK1-NEXT: [[TMP2:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
+// CHECK1-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP2]], 0
+// CHECK1-NEXT: br i1 [[LOADEDV]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
+// CHECK1: omp_if.then:
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208.omp_outlined)
+// CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
+// CHECK1: omp_if.else:
+// CHECK1-NEXT: call void @__kmpc_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTTHREADID_TEMP_]], align 4
+// CHECK1-NEXT: store i32 0, ptr [[DOTBOUND_ZERO_ADDR]], align 4
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208.omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTBOUND_ZERO_ADDR]]) #[[ATTR3]]
+// CHECK1-NEXT: call void @__kmpc_end_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK1-NEXT: br label [[OMP_IF_END]]
+// CHECK1: omp_if.end:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z38conditional_target_parallel_with_limiti
+// CHECK1-SAME: (i32 noundef signext [[COND:%.*]]) #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[COND_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 8
+// CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_24:%.*]], align 1
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store i32 [[COND]], ptr [[COND_ADDR]], align 4
+// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[COND_ADDR]], align 4
+// CHECK1-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP1]], 0
+// CHECK1-NEXT: [[STOREDV:%.*]] = zext i1 [[TOBOOL]] to i8
+// CHECK1-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK1-NEXT: [[TMP2:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK1-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP2]], 0
+// CHECK1-NEXT: [[STOREDV1:%.*]] = zext i1 [[LOADEDV]] to i8
+// CHECK1-NEXT: store i8 [[STOREDV1]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1
+// CHECK1-NEXT: [[TMP3:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store i64 [[TMP3]], ptr [[TMP4]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store i64 [[TMP3]], ptr [[TMP5]], align 8
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
+// CHECK1-NEXT: store ptr null, ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
+// CHECK1-NEXT: store ptr null, ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
+// CHECK1-NEXT: store ptr null, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_24]], ptr [[AGG_CAPTURED]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP13:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK1-NEXT: [[LOADEDV2:%.*]] = icmp ne i8 [[TMP13]], 0
+// CHECK1-NEXT: [[STOREDV3:%.*]] = zext i1 [[LOADEDV2]] to i8
+// CHECK1-NEXT: store i8 [[STOREDV3]], ptr [[TMP12]], align 1
+// CHECK1-NEXT: [[TMP14:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i64 88, i64 1, ptr @.omp_task_entry..89)
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_26:%.*]], ptr [[TMP14]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP15]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP17:%.*]] = load ptr, ptr [[TMP16]], align 8
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 1 [[TMP17]], ptr align 1 [[AGG_CAPTURED]], i64 1, i1 false)
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_26]], ptr [[TMP14]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_25:%.*]], ptr [[TMP18]], i32 0, i32 0
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP19]], ptr align 8 [[TMP10]], i64 16, i1 false)
+// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_25]], ptr [[TMP18]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP20]], ptr align 8 [[TMP11]], i64 16, i1 false)
+// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_25]], ptr [[TMP18]], i32 0, i32 2
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP21]], ptr align 8 @.offload_sizes.85, i64 16, i1 false)
+// CHECK1-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP14]])
+// CHECK1-NEXT: [[TMP22:%.*]] = call i32 @.omp_task_entry..89(i32 [[TMP0]], ptr [[TMP14]]) #[[ATTR3]]
+// CHECK1-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP14]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213
+// CHECK1-SAME: (i64 noundef [[DOTCAPTURE_EXPR_:%.*]], ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK1-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
+// CHECK1-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP1]], 0
+// CHECK1-NEXT: br i1 [[LOADEDV]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
+// CHECK1: omp_if.then:
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213.omp_outlined)
+// CHECK1-NEXT: br label [[OMP_IF_END:%.*]]
+// CHECK1: omp_if.else:
+// CHECK1-NEXT: call void @__kmpc_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTTHREADID_TEMP_]], align 4
+// CHECK1-NEXT: store i32 0, ptr [[DOTBOUND_ZERO_ADDR]], align 4
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213.omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTBOUND_ZERO_ADDR]]) #[[ATTR3]]
+// CHECK1-NEXT: call void @__kmpc_end_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK1-NEXT: br label [[OMP_IF_END]]
+// CHECK1: omp_if.end:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map..88
+// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_25:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_25]], ptr [[TMP4]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_25]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry..89
+// CHECK1-SAME: (i32 noundef signext [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED_I:%.*]] = alloca i64, align 8
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_26:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_26]], ptr [[TMP3]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META195:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META198:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META200:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META202:![0-9]+]])
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META204:![0-9]+]]
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: store ptr @.omp_task_privates_map..88, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META205:![0-9]+]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META204]]
+// CHECK1-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK1-NEXT: [[TMP16:%.*]] = load i8, ptr [[TMP9]], align 1
+// CHECK1-NEXT: [[LOADEDV_I:%.*]] = icmp ne i8 [[TMP16]], 0
+// CHECK1-NEXT: [[TMP17:%.*]] = select i1 [[LOADEDV_I]], i32 16, i32 1
+// CHECK1-NEXT: [[TMP18:%.*]] = select i1 [[LOADEDV_I]], i32 4, i32 1
+// CHECK1-NEXT: [[TMP19:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP17]], 0
+// CHECK1-NEXT: [[TMP20:%.*]] = insertvalue [3 x i32] [[TMP19]], i32 [[TMP18]], 1
+// CHECK1-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK1-NEXT: store i32 2, ptr [[TMP21]], align 4, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP22]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP23]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP24]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.86, ptr [[TMP25]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP26]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP27]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP28]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP29]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP30]], align 4, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [[TMP20]], ptr [[TMP31]], align 4, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP32]], align 4, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 [[TMP17]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK1-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
+// CHECK1-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__87_EXIT:%.*]]
+// CHECK1: omp_offload.failed.i:
+// CHECK1-NEXT: [[TMP35:%.*]] = load i8, ptr [[TMP9]], align 1
+// CHECK1-NEXT: [[LOADEDV3_I:%.*]] = icmp ne i8 [[TMP35]], 0
+// CHECK1-NEXT: [[STOREDV_I:%.*]] = zext i1 [[LOADEDV3_I]] to i8
+// CHECK1-NEXT: store i8 [[STOREDV_I]], ptr [[DOTCAPTURE_EXPR__CASTED_I]], align 1, !noalias [[META204]]
+// CHECK1-NEXT: [[TMP36:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED_I]], align 8, !noalias [[META204]]
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213(i64 [[TMP36]], ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[DOTOMP_OUTLINED__87_EXIT]]
+// CHECK1: .omp_outlined..87.exit:
+// CHECK1-NEXT: ret i32 0
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z24disabled_target_parallelv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.90, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.91, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24disabled_target_parallelv_l218.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24disabled_target_parallelv_l218(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24disabled_target_parallelv_l218
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK1-NEXT: call void @__kmpc_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTTHREADID_TEMP_]], align 4
+// CHECK1-NEXT: store i32 0, ptr [[DOTBOUND_ZERO_ADDR]], align 4
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24disabled_target_parallelv_l218.omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTBOUND_ZERO_ADDR]]) #[[ATTR3]]
+// CHECK1-NEXT: call void @__kmpc_end_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24disabled_target_parallelv_l218.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z23enabled_target_parallelv
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP0]], align 8
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK1-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK1-NEXT: store ptr @.offload_sizes.92, ptr [[TMP9]], align 8
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.93, ptr [[TMP10]], align 8
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
+// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
+// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 16, i32 4, i32 0], ptr [[TMP16]], align 4
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23enabled_target_parallelv_l223.region_id, ptr [[KERNEL_ARGS]])
+// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK1: omp_offload.failed:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23enabled_target_parallelv_l223(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK1: omp_offload.cont:
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23enabled_target_parallelv_l223
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23enabled_target_parallelv_l223.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23enabled_target_parallelv_l223.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z24extra_limit_dimensions_1v
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_27:%.*]], align 1
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i64 64, i64 1, ptr @.omp_task_entry..98)
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_29:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_29]], ptr [[TMP6]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_28:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP9]], ptr align 8 [[TMP4]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_28]], ptr [[TMP8]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP10]], ptr align 8 [[TMP5]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_28]], ptr [[TMP8]], i32 0, i32 2
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP11]], ptr align 8 @.offload_sizes.94, i64 8, i1 false)
+// CHECK1-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry..98(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK1-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_1v_l230
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_1v_l230.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_1v_l230.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map..97
+// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_28:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_28]], ptr [[TMP4]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_28]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry..98
+// CHECK1-SAME: (i32 noundef signext [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_29:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_29]], ptr [[TMP3]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META206:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META209:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META211:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META213:![0-9]+]])
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META215:![0-9]+]]
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: store ptr @.omp_task_privates_map..97, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META216:![0-9]+]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META215]]
+// CHECK1-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK1-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.95, ptr [[TMP20]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP21]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP22]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP25]], align 4, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 16, i32 1, i32 0], ptr [[TMP26]], align 4, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META215]]
+// CHECK1-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_1v_l230.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK1-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK1-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__96_EXIT:%.*]]
+// CHECK1: omp_offload.failed.i:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_1v_l230(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[DOTOMP_OUTLINED__96_EXIT]]
+// CHECK1: .omp_outlined..96.exit:
+// CHECK1-NEXT: ret i32 0
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z24extra_limit_dimensions_2v
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_30:%.*]], align 1
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i64 64, i64 1, ptr @.omp_task_entry..103)
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_32:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_32]], ptr [[TMP6]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_31:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP9]], ptr align 8 [[TMP4]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_31]], ptr [[TMP8]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP10]], ptr align 8 [[TMP5]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_31]], ptr [[TMP8]], i32 0, i32 2
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP11]], ptr align 8 @.offload_sizes.99, i64 8, i1 false)
+// CHECK1-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry..103(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK1-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_2v_l235
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_2v_l235.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_2v_l235.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map..102
+// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_31:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_31]], ptr [[TMP4]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_31]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry..103
+// CHECK1-SAME: (i32 noundef signext [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_32:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_32]], ptr [[TMP3]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META217:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META220:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META222:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META224:![0-9]+]])
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META226:![0-9]+]]
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: store ptr @.omp_task_privates_map..102, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META227:![0-9]+]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META226]]
+// CHECK1-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK1-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.100, ptr [[TMP20]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP21]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP22]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP25]], align 4, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 16, i32 1, i32 0], ptr [[TMP26]], align 4, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META226]]
+// CHECK1-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_2v_l235.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK1-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK1-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__101_EXIT:%.*]]
+// CHECK1: omp_offload.failed.i:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_2v_l235(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[DOTOMP_OUTLINED__101_EXIT]]
+// CHECK1: .omp_outlined..101.exit:
+// CHECK1-NEXT: ret i32 0
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z24extra_limit_dimensions_3v
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_33:%.*]], align 1
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i64 64, i64 1, ptr @.omp_task_entry..108)
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_35:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_35]], ptr [[TMP6]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_34:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP9]], ptr align 8 [[TMP4]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_34]], ptr [[TMP8]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP10]], ptr align 8 [[TMP5]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_34]], ptr [[TMP8]], i32 0, i32 2
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP11]], ptr align 8 @.offload_sizes.104, i64 8, i1 false)
+// CHECK1-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry..108(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK1-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_3v_l240
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_3v_l240.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_3v_l240.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map..107
+// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_34:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_34]], ptr [[TMP4]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_34]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry..108
+// CHECK1-SAME: (i32 noundef signext [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_35:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_35]], ptr [[TMP3]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META228:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META231:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META233:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META235:![0-9]+]])
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META237:![0-9]+]]
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: store ptr @.omp_task_privates_map..107, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META238:![0-9]+]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META237]]
+// CHECK1-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK1-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.105, ptr [[TMP20]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP21]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP22]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP25]], align 4, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 16, i32 1, i32 1], ptr [[TMP26]], align 4, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META237]]
+// CHECK1-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_3v_l240.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK1-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK1-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__106_EXIT:%.*]]
+// CHECK1: omp_offload.failed.i:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_3v_l240(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[DOTOMP_OUTLINED__106_EXIT]]
+// CHECK1: .omp_outlined..106.exit:
+// CHECK1-NEXT: ret i32 0
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z24extra_limit_dimensions_4v
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_36:%.*]], align 1
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i64 64, i64 1, ptr @.omp_task_entry..113)
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_38:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_38]], ptr [[TMP6]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_37:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP9]], ptr align 8 [[TMP4]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_37]], ptr [[TMP8]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP10]], ptr align 8 [[TMP5]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_37]], ptr [[TMP8]], i32 0, i32 2
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP11]], ptr align 8 @.offload_sizes.109, i64 8, i1 false)
+// CHECK1-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry..113(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK1-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_4v_l245
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_4v_l245.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_4v_l245.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map..112
+// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_37:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_37]], ptr [[TMP4]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_37]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry..113
+// CHECK1-SAME: (i32 noundef signext [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_38:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_38]], ptr [[TMP3]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META239:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META242:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META244:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META246:![0-9]+]])
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META248:![0-9]+]]
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: store ptr @.omp_task_privates_map..112, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META249:![0-9]+]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META248]]
+// CHECK1-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK1-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.110, ptr [[TMP20]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP21]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP22]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP25]], align 4, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 16, i32 1, i32 1], ptr [[TMP26]], align 4, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META248]]
+// CHECK1-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_4v_l245.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK1-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK1-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__111_EXIT:%.*]]
+// CHECK1: omp_offload.failed.i:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_4v_l245(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[DOTOMP_OUTLINED__111_EXIT]]
+// CHECK1: .omp_outlined..111.exit:
+// CHECK1-NEXT: ret i32 0
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@_Z24extra_limit_dimensions_5v
+// CHECK1-SAME: () #[[ATTR0]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
+// CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_39:%.*]], align 1
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
+// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
+// CHECK1-NEXT: store ptr null, ptr [[TMP3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i64 64, i64 1, ptr @.omp_task_entry..118)
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_41:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_41]], ptr [[TMP6]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_40:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP9]], ptr align 8 [[TMP4]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_40]], ptr [[TMP8]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP10]], ptr align 8 [[TMP5]], i64 8, i1 false)
+// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_40]], ptr [[TMP8]], i32 0, i32 2
+// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP11]], ptr align 8 @.offload_sizes.114, i64 8, i1 false)
+// CHECK1-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry..118(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK1-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_5v_l250
+// CHECK1-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK1-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
+// CHECK1-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_5v_l250.omp_outlined)
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_5v_l250.omp_outlined
+// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map..117
+// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 8
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_40:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_40]], ptr [[TMP4]], i32 0, i32 1
+// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR2]], align 8
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 8
+// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_40]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR3]], align 8
+// CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 8
+// CHECK1-NEXT: ret void
+//
+//
+// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry..118
+// CHECK1-SAME: (i32 noundef signext [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK1-NEXT: entry:
+// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
+// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
+// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_41:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
+// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_41]], ptr [[TMP3]], i32 0, i32 1
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META250:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META253:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META255:![0-9]+]])
+// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META257:![0-9]+]])
+// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META259:![0-9]+]]
+// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: store ptr @.omp_task_privates_map..117, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META260:![0-9]+]]
+// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META259]]
+// CHECK1-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK1-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK1-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK1-NEXT: store ptr @.offload_maptypes.115, ptr [[TMP20]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK1-NEXT: store ptr null, ptr [[TMP21]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK1-NEXT: store ptr null, ptr [[TMP22]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK1-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK1-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP25]], align 4, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK1-NEXT: store [3 x i32] [i32 16, i32 4, i32 1], ptr [[TMP26]], align 4, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK1-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META259]]
+// CHECK1-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_5v_l250.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK1-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK1-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__116_EXIT:%.*]]
+// CHECK1: omp_offload.failed.i:
+// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_5v_l250(ptr null) #[[ATTR3]]
+// CHECK1-NEXT: br label [[DOTOMP_OUTLINED__116_EXIT]]
+// CHECK1: .omp_outlined..116.exit:
+// CHECK1-NEXT: ret i32 0
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z25one_dimensional_modifiersv
+// CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 3, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 16, i32 0, i32 0], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1:[0-9]+]], i64 -1, i32 3, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z25one_dimensional_modifiersv_l47.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z25one_dimensional_modifiersv_l47(ptr null) #[[ATTR3:[0-9]+]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z25one_dimensional_modifiersv_l47
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR1:[0-9]+]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 3, i32 16)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z25one_dimensional_modifiersv_l47.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z25one_dimensional_modifiersv_l47.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2:[0-9]+]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z22target_without_clausesv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.1, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_without_clausesv_l52.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_without_clausesv_l52(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_without_clausesv_l52
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR2]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z27target_teams_num_teams_onlyv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.3, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 2, i32 3, i32 4], ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 2, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27target_teams_num_teams_onlyv_l58.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27target_teams_num_teams_onlyv_l58(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27target_teams_num_teams_onlyv_l58
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR4:[0-9]+]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 2, i32 0)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27target_teams_num_teams_onlyv_l58.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27target_teams_num_teams_onlyv_l58.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z30target_teams_num_teams_dynamicii
+// CHECK3-SAME: (i32 noundef [[TEAMS_X:%.*]], i32 noundef [[TEAMS_Y:%.*]]) #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[TEAMS_X_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[TEAMS_Y_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED2:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: store i32 [[TEAMS_X]], ptr [[TEAMS_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TEAMS_Y]], ptr [[TEAMS_Y_ADDR]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = load i32, ptr [[TEAMS_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[TEAMS_Y_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR__CASTED2]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED2]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 [[TMP3]], ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 [[TMP5]], ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 [[TMP5]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr null, ptr [[TMP13]], align 4
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: [[TMP19:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP17]], 0
+// CHECK3-NEXT: [[TMP20:%.*]] = insertvalue [3 x i32] [[TMP19]], i32 [[TMP18]], 1
+// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP21]], align 4
+// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 3, ptr [[TMP22]], align 4
+// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP15]], ptr [[TMP23]], align 4
+// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP16]], ptr [[TMP24]], align 4
+// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.5, ptr [[TMP25]], align 4
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP26]], align 4
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP27]], align 4
+// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP28]], align 4
+// CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP29]], align 8
+// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP30]], align 8
+// CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [[TMP20]], ptr [[TMP31]], align 4
+// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP32]], align 4
+// CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP33]], align 4
+// CHECK3-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 [[TMP17]], i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_num_teams_dynamicii_l63.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
+// CHECK3-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_num_teams_dynamicii_l63(i32 [[TMP3]], i32 [[TMP5]], ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_num_teams_dynamicii_l63
+// CHECK3-SAME: (i32 noundef [[DOTCAPTURE_EXPR_:%.*]], i32 noundef [[DOTCAPTURE_EXPR_1:%.*]], ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR2]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
+// CHECK3-NEXT: store i32 [[DOTCAPTURE_EXPR_1]], ptr [[DOTCAPTURE_EXPR__ADDR2]], align 4
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP1]], i32 0)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_num_teams_dynamicii_l63.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_num_teams_dynamicii_l63.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z30target_teams_thread_limit_onlyv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.7, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 32, i32 8, i32 4], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 32, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_thread_limit_onlyv_l68.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_thread_limit_onlyv_l68(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_thread_limit_onlyv_l68
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR5:[0-9]+]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_thread_limit_onlyv_l68.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_thread_limit_onlyv_l68.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z19target_teams_limitsv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.9, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.10, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 3, i32 2, i32 0], ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 16, i32 8, i32 0], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 3, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_teams_limitsv_l73.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_teams_limitsv_l73(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_teams_limitsv_l73
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR1]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 3, i32 16)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_teams_limitsv_l73.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_teams_limitsv_l73.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z24target_thread_limit_onlyv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 1
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 36, i32 1, ptr @.omp_task_entry.)
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP6]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP9]], ptr align 4 @.offload_sizes.11, i32 8, i1 false)
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T]], ptr [[TMP8]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP10]], ptr align 4 [[TMP4]], i32 4, i1 false)
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T]], ptr [[TMP8]], i32 0, i32 2
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP11]], ptr align 4 [[TMP5]], i32 4, i1 false)
+// CHECK3-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry.(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK3-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_thread_limit_onlyv_l78
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR5]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_privates_map.
+// CHECK3-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6:[0-9]+]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T]], ptr [[TMP4]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_entry.
+// CHECK3-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7:[0-9]+]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP3]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META83:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META86:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META88:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META90:![0-9]+]])
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META92:![0-9]+]]
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: store ptr @.omp_task_privates_map., ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META93:![0-9]+]]
+// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK3-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.12, ptr [[TMP20]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP21]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP22]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP25]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 32, i32 8, i32 0], ptr [[TMP26]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META92]]
+// CHECK3-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 32, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_thread_limit_onlyv_l78.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK3-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK3-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__EXIT:%.*]]
+// CHECK3: omp_offload.failed.i:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_thread_limit_onlyv_l78(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[DOTOMP_OUTLINED__EXIT]]
+// CHECK3: .omp_outlined..exit:
+// CHECK3-NEXT: ret i32 0
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z22target_nested_parallelii
+// CHECK3-SAME: (i32 noundef [[THREADS_X:%.*]], i32 noundef [[THREADS_Y:%.*]]) #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[THREADS_X_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[THREADS_Y_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[THREADS_X_CASTED:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[THREADS_Y_CASTED:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4
+// CHECK3-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_0:%.*]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store i32 [[THREADS_X]], ptr [[THREADS_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[THREADS_Y]], ptr [[THREADS_Y_ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[THREADS_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP1]], ptr [[THREADS_X_CASTED]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[THREADS_X_CASTED]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[THREADS_Y_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP3]], ptr [[THREADS_Y_CASTED]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[THREADS_Y_CASTED]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
+// CHECK3-NEXT: store ptr null, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr null, ptr [[TMP13]], align 4
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_0]], ptr [[AGG_CAPTURED]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[THREADS_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP17]], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_0]], ptr [[AGG_CAPTURED]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[THREADS_Y_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP19]], ptr [[TMP18]], align 4
+// CHECK3-NEXT: [[TMP20:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 76, i32 8, ptr @.omp_task_entry..17)
+// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2:%.*]], ptr [[TMP20]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP21]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP23:%.*]] = load ptr, ptr [[TMP22]], align 4
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP23]], ptr align 4 [[AGG_CAPTURED]], i32 8, i1 false)
+// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2]], ptr [[TMP20]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1:%.*]], ptr [[TMP24]], i32 0, i32 0
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP25]], ptr align 4 @.offload_sizes.13, i32 24, i1 false)
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1]], ptr [[TMP24]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP27:%.*]] = load i32, ptr [[THREADS_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP27]], ptr [[TMP26]], align 4
+// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1]], ptr [[TMP24]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP29:%.*]] = load i32, ptr [[THREADS_Y_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP29]], ptr [[TMP28]], align 4
+// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1]], ptr [[TMP24]], i32 0, i32 3
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP30]], ptr align 4 [[TMP14]], i32 12, i1 false)
+// CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1]], ptr [[TMP24]], i32 0, i32 4
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP31]], ptr align 4 [[TMP15]], i32 12, i1 false)
+// CHECK3-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP20]])
+// CHECK3-NEXT: [[TMP32:%.*]] = call i32 @.omp_task_entry..17(i32 [[TMP0]], ptr [[TMP20]]) #[[ATTR3]]
+// CHECK3-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP20]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_nested_parallelii_l83
+// CHECK3-SAME: (i32 noundef [[THREADS_X:%.*]], i32 noundef [[THREADS_Y:%.*]], ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR9:[0-9]+]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[THREADS_X_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[THREADS_Y_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store i32 [[THREADS_X]], ptr [[THREADS_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[THREADS_Y]], ptr [[THREADS_Y_ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[THREADS_X_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP1]])
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_nested_parallelii_l83.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_nested_parallelii_l83.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10:[0-9]+]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_privates_map..16
+// CHECK3-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]], ptr noalias noundef [[TMP4:%.*]], ptr noalias noundef [[TMP5:%.*]]) #[[ATTR6]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR4:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR5:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[DOTADDR4]], align 4
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[DOTADDR5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR5]], align 4
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1]], ptr [[TMP6]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1]], ptr [[TMP6]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP11]], ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1]], ptr [[TMP6]], i32 0, i32 3
+// CHECK3-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: store ptr [[TMP13]], ptr [[TMP14]], align 4
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_1]], ptr [[TMP6]], i32 0, i32 4
+// CHECK3-NEXT: [[TMP16:%.*]] = load ptr, ptr [[DOTADDR4]], align 4
+// CHECK3-NEXT: store ptr [[TMP15]], ptr [[TMP16]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_entry..17
+// CHECK3-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR3_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR4_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[THREADS_X_CASTED_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[THREADS_Y_CASTED_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2]], ptr [[TMP3]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META94:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META97:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META99:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META101:![0-9]+]])
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META103:![0-9]+]]
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: store ptr @.omp_task_privates_map..16, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR3_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR4_I]]) #[[ATTR3]], !inline_history [[META104:![0-9]+]]
+// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP15:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR3_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP16:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR4_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP17]], i32 32)
+// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP13]], align 4
+// CHECK3-NEXT: [[TMP20:%.*]] = icmp ult i32 32, [[TMP18]]
+// CHECK3-NEXT: [[TMP21:%.*]] = select i1 [[TMP20]], i32 32, i32 [[TMP18]]
+// CHECK3-NEXT: [[TMP22:%.*]] = icmp ult i32 8, [[TMP19]]
+// CHECK3-NEXT: [[TMP23:%.*]] = select i1 [[TMP22]], i32 8, i32 [[TMP19]]
+// CHECK3-NEXT: [[TMP24:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP21]], 0
+// CHECK3-NEXT: [[TMP25:%.*]] = insertvalue [3 x i32] [[TMP24]], i32 [[TMP23]], 1
+// CHECK3-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK3-NEXT: store i32 3, ptr [[TMP26]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP27]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP15]], ptr [[TMP28]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK3-NEXT: store ptr [[TMP16]], ptr [[TMP29]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.14, ptr [[TMP30]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP31]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP32]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP33]], align 8, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP34]], align 8, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP35]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [[TMP25]], ptr [[TMP36]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP37]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 [[TMP21]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_nested_parallelii_l83.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK3-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0
+// CHECK3-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__15_EXIT:%.*]]
+// CHECK3: omp_offload.failed.i:
+// CHECK3-NEXT: [[TMP40:%.*]] = load i32, ptr [[TMP12]], align 4
+// CHECK3-NEXT: store i32 [[TMP40]], ptr [[THREADS_X_CASTED_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP41:%.*]] = load i32, ptr [[THREADS_X_CASTED_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP42:%.*]] = load i32, ptr [[TMP13]], align 4
+// CHECK3-NEXT: store i32 [[TMP42]], ptr [[THREADS_Y_CASTED_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: [[TMP43:%.*]] = load i32, ptr [[THREADS_Y_CASTED_I]], align 4, !noalias [[META103]]
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_nested_parallelii_l83(i32 [[TMP41]], i32 [[TMP43]], ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[DOTOMP_OUTLINED__15_EXIT]]
+// CHECK3: .omp_outlined..15.exit:
+// CHECK3-NEXT: ret i32 0
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z28target_nested_teams_parallelv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.18, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.19, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 32, i32 8, i32 0], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 32, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR11:[0-9]+]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89.omp_outlined.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89.omp_outlined.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z41nested_teams_limit_overrides_target_limitv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_3:%.*]], align 1
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 36, i32 1, ptr @.omp_task_entry..24)
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_5:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_5]], ptr [[TMP6]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_4:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP9]], ptr align 4 @.offload_sizes.20, i32 8, i1 false)
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_4]], ptr [[TMP8]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP10]], ptr align 4 [[TMP4]], i32 4, i1 false)
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_4]], ptr [[TMP8]], i32 0, i32 2
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP11]], ptr align 4 [[TMP5]], i32 4, i1 false)
+// CHECK3-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry..24(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK3-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41nested_teams_limit_overrides_target_limitv_l96
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR5]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41nested_teams_limit_overrides_target_limitv_l96.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41nested_teams_limit_overrides_target_limitv_l96.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_privates_map..23
+// CHECK3-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_4:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_4]], ptr [[TMP4]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_4]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_entry..24
+// CHECK3-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_5:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_5]], ptr [[TMP3]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META105:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META108:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META110:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META112:![0-9]+]])
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META114:![0-9]+]]
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: store ptr @.omp_task_privates_map..23, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META115:![0-9]+]]
+// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 64)
+// CHECK3-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.21, ptr [[TMP20]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP21]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP22]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP25]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 32, i32 8, i32 0], ptr [[TMP26]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META114]]
+// CHECK3-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 32, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41nested_teams_limit_overrides_target_limitv_l96.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK3-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK3-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__22_EXIT:%.*]]
+// CHECK3: omp_offload.failed.i:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41nested_teams_limit_overrides_target_limitv_l96(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[DOTOMP_OUTLINED__22_EXIT]]
+// CHECK3: .omp_outlined..22.exit:
+// CHECK3-NEXT: ret i32 0
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z23target_teams_distributev
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.25, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.26, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 4, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 32, i32 8, i32 0], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 32, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_teams_distributev_l102.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_teams_distributev_l102(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_teams_distributev_l102
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR5]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_teams_distributev_l102.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_teams_distributev_l102.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK3: cond.true:
+// CHECK3-NEXT: br label [[COND_END:%.*]]
+// CHECK3: cond.false:
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: br label [[COND_END]]
+// CHECK3: cond.end:
+// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK3: omp.inner.for.cond:
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK3: omp.inner.for.body:
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK3: omp.body.continue:
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK3: omp.inner.for.inc:
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK3: omp.loop.exit:
+// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z36target_teams_distribute_parallel_forv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.27, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.28, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 4, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 16, i32 4, i32 0], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR11]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: store i32 3, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK3: cond.true:
+// CHECK3-NEXT: br label [[COND_END:%.*]]
+// CHECK3: cond.false:
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: br label [[COND_END]]
+// CHECK3: cond.end:
+// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK3: omp.inner.for.cond:
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK3: omp.inner.for.body:
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16)
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107.omp_outlined.omp_outlined, i32 [[TMP7]], i32 [[TMP8]])
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK3: omp.inner.for.inc:
+// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
+// CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK3: omp.loop.exit:
+// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107.omp_outlined.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
+// CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 3
+// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK3: cond.true:
+// CHECK3-NEXT: br label [[COND_END:%.*]]
+// CHECK3: cond.false:
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: br label [[COND_END]]
+// CHECK3: cond.end:
+// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
+// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK3: omp.inner.for.cond:
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
+// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK3: omp.inner.for.body:
+// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
+// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK3: omp.body.continue:
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK3: omp.inner.for.inc:
+// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1
+// CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK3: omp.loop.exit:
+// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z41target_teams_distribute_parallel_for_simdv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.29, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.30, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 4, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 16, i32 4, i32 0], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR11]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: store i32 3, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK3: cond.true:
+// CHECK3-NEXT: br label [[COND_END:%.*]]
+// CHECK3: cond.false:
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: br label [[COND_END]]
+// CHECK3: cond.end:
+// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK3: omp.inner.for.cond:
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP116:![0-9]+]]
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK3: omp.inner.for.body:
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16), !llvm.access.group [[ACC_GRP116]]
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112.omp_outlined.omp_outlined, i32 [[TMP7]], i32 [[TMP8]]), !llvm.access.group [[ACC_GRP116]]
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK3: omp.inner.for.inc:
+// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
+// CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP116]]
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP117:![0-9]+]]
+// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK3: omp.loop.exit:
+// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
+// CHECK3-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK3: .omp.final.then:
+// CHECK3-NEXT: store i32 4, ptr [[I]], align 4
+// CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
+// CHECK3: .omp.final.done:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112.omp_outlined.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
+// CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 3
+// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK3: cond.true:
+// CHECK3-NEXT: br label [[COND_END:%.*]]
+// CHECK3: cond.false:
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: br label [[COND_END]]
+// CHECK3: cond.end:
+// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
+// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK3: omp.inner.for.cond:
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120:![0-9]+]]
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP120]]
+// CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
+// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK3: omp.inner.for.body:
+// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120]]
+// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
+// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP120]]
+// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK3: omp.body.continue:
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK3: omp.inner.for.inc:
+// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120]]
+// CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1
+// CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP120]]
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP121:![0-9]+]]
+// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK3: omp.loop.exit:
+// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
+// CHECK3-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK3: .omp.final.then:
+// CHECK3-NEXT: store i32 4, ptr [[I]], align 4
+// CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
+// CHECK3: .omp.final.done:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z28target_teams_distribute_simdv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.31, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.32, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 4, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 3, i32 2, i32 0], ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 3, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_teams_distribute_simdv_l117.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_teams_distribute_simdv_l117(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_teams_distribute_simdv_l117
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR12:[0-9]+]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 3, i32 32)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_teams_distribute_simdv_l117.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_teams_distribute_simdv_l117.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK3: cond.true:
+// CHECK3-NEXT: br label [[COND_END:%.*]]
+// CHECK3: cond.false:
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: br label [[COND_END]]
+// CHECK3: cond.end:
+// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK3: omp.inner.for.cond:
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123:![0-9]+]]
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK3: omp.inner.for.body:
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK3: omp.body.continue:
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK3: omp.inner.for.inc:
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP123]]
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP124:![0-9]+]]
+// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK3: omp.loop.exit:
+// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = icmp ne i32 [[TMP9]], 0
+// CHECK3-NEXT: br i1 [[TMP10]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK3: .omp.final.then:
+// CHECK3-NEXT: store i32 4, ptr [[I]], align 4
+// CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
+// CHECK3: .omp.final.done:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z17target_teams_loopv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.33, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.34, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 4, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 3, i32 2, i32 0], ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 32, i32 8, i32 0], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 3, i32 32, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR13:[0-9]+]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 3, i32 32)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: store i32 3, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK3: cond.true:
+// CHECK3-NEXT: br label [[COND_END:%.*]]
+// CHECK3: cond.false:
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: br label [[COND_END]]
+// CHECK3: cond.end:
+// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK3: omp.inner.for.cond:
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK3: omp.inner.for.body:
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122.omp_outlined.omp_outlined, i32 [[TMP7]], i32 [[TMP8]])
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK3: omp.inner.for.inc:
+// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
+// CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK3: omp.loop.exit:
+// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122.omp_outlined.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
+// CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 3
+// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK3: cond.true:
+// CHECK3-NEXT: br label [[COND_END:%.*]]
+// CHECK3: cond.false:
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: br label [[COND_END]]
+// CHECK3: cond.end:
+// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
+// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK3: omp.inner.for.cond:
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
+// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK3: omp.inner.for.body:
+// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
+// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK3: omp.body.continue:
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK3: omp.inner.for.inc:
+// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1
+// CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK3: omp.loop.exit:
+// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z33target_parallel_thread_limit_onlyv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_6:%.*]], align 1
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 36, i32 1, ptr @.omp_task_entry..39)
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_8:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_8]], ptr [[TMP6]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_7:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP9]], ptr align 4 @.offload_sizes.35, i32 8, i1 false)
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_7]], ptr [[TMP8]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP10]], ptr align 4 [[TMP4]], i32 4, i1 false)
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_7]], ptr [[TMP8]], i32 0, i32 2
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP11]], ptr align 4 [[TMP5]], i32 4, i1 false)
+// CHECK3-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry..39(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK3-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z33target_parallel_thread_limit_onlyv_l128
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR9]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z33target_parallel_thread_limit_onlyv_l128.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z33target_parallel_thread_limit_onlyv_l128.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_privates_map..38
+// CHECK3-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_7:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_7]], ptr [[TMP4]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_7]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_entry..39
+// CHECK3-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_8:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_8]], ptr [[TMP3]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META126:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META129:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META131:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META133:![0-9]+]])
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META135:![0-9]+]]
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: store ptr @.omp_task_privates_map..38, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META136:![0-9]+]]
+// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK3-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.36, ptr [[TMP20]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP21]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP22]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP25]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 32, i32 8, i32 0], ptr [[TMP26]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META135]]
+// CHECK3-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 32, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z33target_parallel_thread_limit_onlyv_l128.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK3-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK3-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__37_EXIT:%.*]]
+// CHECK3: omp_offload.failed.i:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z33target_parallel_thread_limit_onlyv_l128(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[DOTOMP_OUTLINED__37_EXIT]]
+// CHECK3: .omp_outlined..37.exit:
+// CHECK3-NEXT: ret i32 0
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z31target_parallel_threads_dynamicii
+// CHECK3-SAME: (i32 noundef [[THREADS_X:%.*]], i32 noundef [[THREADS_Y:%.*]]) #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[THREADS_X_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[THREADS_Y_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED2:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: store i32 [[THREADS_X]], ptr [[THREADS_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[THREADS_Y]], ptr [[THREADS_Y_ADDR]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = load i32, ptr [[THREADS_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[THREADS_Y_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR__CASTED2]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED2]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 [[TMP3]], ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 [[TMP5]], ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 [[TMP5]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr null, ptr [[TMP13]], align 4
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: [[TMP19:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP17]], 0
+// CHECK3-NEXT: [[TMP20:%.*]] = insertvalue [3 x i32] [[TMP19]], i32 [[TMP18]], 1
+// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP21]], align 4
+// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 3, ptr [[TMP22]], align 4
+// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP15]], ptr [[TMP23]], align 4
+// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP16]], ptr [[TMP24]], align 4
+// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.40, ptr [[TMP25]], align 4
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.41, ptr [[TMP26]], align 4
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP27]], align 4
+// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP28]], align 4
+// CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP29]], align 8
+// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP30]], align 8
+// CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP31]], align 4
+// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [[TMP20]], ptr [[TMP32]], align 4
+// CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP33]], align 4
+// CHECK3-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 [[TMP17]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z31target_parallel_threads_dynamicii_l133.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
+// CHECK3-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z31target_parallel_threads_dynamicii_l133(i32 [[TMP3]], i32 [[TMP5]], ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z31target_parallel_threads_dynamicii_l133
+// CHECK3-SAME: (i32 noundef [[DOTCAPTURE_EXPR_:%.*]], i32 noundef [[DOTCAPTURE_EXPR_1:%.*]], ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR14:[0-9]+]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
+// CHECK3-NEXT: store i32 [[DOTCAPTURE_EXPR_1]], ptr [[DOTCAPTURE_EXPR__ADDR2]], align 4
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP1]])
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z31target_parallel_threads_dynamicii_l133.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z31target_parallel_threads_dynamicii_l133.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z36target_parallel_one_thread_dimensionv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.42, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.43, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 16, i32 0, i32 0], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_parallel_one_thread_dimensionv_l138.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_parallel_one_thread_dimensionv_l138(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_parallel_one_thread_dimensionv_l138
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15:[0-9]+]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_parallel_one_thread_dimensionv_l138.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_parallel_one_thread_dimensionv_l138.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z23target_parallel_threadsv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.44, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.45, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 8, i32 4, i32 2], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 8, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_threadsv_l143.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_threadsv_l143(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_threadsv_l143
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR16:[0-9]+]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 8)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_threadsv_l143.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_threadsv_l143.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z19target_parallel_forv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.46, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.47, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 16, i32 4, i32 0], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_parallel_forv_l148.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_parallel_forv_l148(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_parallel_forv_l148
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_parallel_forv_l148.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_parallel_forv_l148.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK3: cond.true:
+// CHECK3-NEXT: br label [[COND_END:%.*]]
+// CHECK3: cond.false:
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: br label [[COND_END]]
+// CHECK3: cond.end:
+// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK3: omp.inner.for.cond:
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK3: omp.inner.for.body:
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK3: omp.body.continue:
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK3: omp.inner.for.inc:
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK3: omp.loop.exit:
+// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z24target_parallel_for_simdv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.48, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.49, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 16, i32 4, i32 0], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_for_simdv_l153.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_for_simdv_l153(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_for_simdv_l153
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_for_simdv_l153.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_for_simdv_l153.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK3: cond.true:
+// CHECK3-NEXT: br label [[COND_END:%.*]]
+// CHECK3: cond.false:
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: br label [[COND_END]]
+// CHECK3: cond.end:
+// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK3: omp.inner.for.cond:
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137:![0-9]+]]
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK3: omp.inner.for.body:
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK3: omp.body.continue:
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK3: omp.inner.for.inc:
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP137]]
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP138:![0-9]+]]
+// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK3: omp.loop.exit:
+// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = icmp ne i32 [[TMP9]], 0
+// CHECK3-NEXT: br i1 [[TMP10]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK3: .omp.final.then:
+// CHECK3-NEXT: store i32 4, ptr [[I]], align 4
+// CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
+// CHECK3: .omp.final.done:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z20target_parallel_loopv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.50, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.51, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 16, i32 4, i32 0], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20target_parallel_loopv_l158.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20target_parallel_loopv_l158(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20target_parallel_loopv_l158
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20target_parallel_loopv_l158.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20target_parallel_loopv_l158.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK3: cond.true:
+// CHECK3-NEXT: br label [[COND_END:%.*]]
+// CHECK3: cond.false:
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: br label [[COND_END]]
+// CHECK3: cond.end:
+// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK3: omp.inner.for.cond:
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK3: omp.inner.for.body:
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK3: omp.body.continue:
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK3: omp.inner.for.inc:
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK3: omp.loop.exit:
+// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z23target_parallel_dynamiciiii
+// CHECK3-SAME: (i32 noundef [[LIMIT_X:%.*]], i32 noundef [[LIMIT_Y:%.*]], i32 noundef [[THREADS_X:%.*]], i32 noundef [[THREADS_Y:%.*]]) #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[LIMIT_X_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[LIMIT_Y_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[THREADS_X_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[THREADS_Y_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED4:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED5:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED6:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x ptr], align 4
+// CHECK3-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_9:%.*]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store i32 [[LIMIT_X]], ptr [[LIMIT_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[LIMIT_Y]], ptr [[LIMIT_Y_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[THREADS_X]], ptr [[THREADS_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[THREADS_Y]], ptr [[THREADS_Y_ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[LIMIT_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[LIMIT_Y_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[THREADS_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[THREADS_Y_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_3]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: store i32 [[TMP7]], ptr [[DOTCAPTURE_EXPR__CASTED4]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED4]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
+// CHECK3-NEXT: store i32 [[TMP9]], ptr [[DOTCAPTURE_EXPR__CASTED5]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED5]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3]], align 4
+// CHECK3-NEXT: store i32 [[TMP11]], ptr [[DOTCAPTURE_EXPR__CASTED6]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED6]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 [[TMP6]], ptr [[TMP13]], align 4
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 [[TMP6]], ptr [[TMP14]], align 4
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 [[TMP8]], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 [[TMP8]], ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
+// CHECK3-NEXT: store ptr null, ptr [[TMP18]], align 4
+// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
+// CHECK3-NEXT: store i32 [[TMP10]], ptr [[TMP19]], align 4
+// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
+// CHECK3-NEXT: store i32 [[TMP10]], ptr [[TMP20]], align 4
+// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr null, ptr [[TMP21]], align 4
+// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
+// CHECK3-NEXT: store i32 [[TMP12]], ptr [[TMP22]], align 4
+// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3
+// CHECK3-NEXT: store i32 [[TMP12]], ptr [[TMP23]], align 4
+// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr null, ptr [[TMP24]], align 4
+// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr null, ptr [[TMP25]], align 4
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr null, ptr [[TMP26]], align 4
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr null, ptr [[TMP27]], align 4
+// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_9]], ptr [[AGG_CAPTURED]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK3-NEXT: store i32 [[TMP31]], ptr [[TMP30]], align 4
+// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_9]], ptr [[AGG_CAPTURED]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: store i32 [[TMP33]], ptr [[TMP32]], align 4
+// CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_9]], ptr [[AGG_CAPTURED]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
+// CHECK3-NEXT: store i32 [[TMP35]], ptr [[TMP34]], align 4
+// CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_9]], ptr [[AGG_CAPTURED]], i32 0, i32 3
+// CHECK3-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3]], align 4
+// CHECK3-NEXT: store i32 [[TMP37]], ptr [[TMP36]], align 4
+// CHECK3-NEXT: [[TMP38:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 100, i32 16, ptr @.omp_task_entry..56)
+// CHECK3-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_11:%.*]], ptr [[TMP38]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP39]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP41:%.*]] = load ptr, ptr [[TMP40]], align 4
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP41]], ptr align 4 [[AGG_CAPTURED]], i32 16, i1 false)
+// CHECK3-NEXT: [[TMP42:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_11]], ptr [[TMP38]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP43:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_10:%.*]], ptr [[TMP42]], i32 0, i32 0
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP43]], ptr align 4 @.offload_sizes.52, i32 40, i1 false)
+// CHECK3-NEXT: [[TMP44:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_10]], ptr [[TMP42]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP44]], ptr align 4 [[TMP28]], i32 20, i1 false)
+// CHECK3-NEXT: [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_10]], ptr [[TMP42]], i32 0, i32 2
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP45]], ptr align 4 [[TMP29]], i32 20, i1 false)
+// CHECK3-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP38]])
+// CHECK3-NEXT: [[TMP46:%.*]] = call i32 @.omp_task_entry..56(i32 [[TMP0]], ptr [[TMP38]]) #[[ATTR3]]
+// CHECK3-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP38]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_dynamiciiii_l165
+// CHECK3-SAME: (i32 noundef [[DOTCAPTURE_EXPR_:%.*]], i32 noundef [[DOTCAPTURE_EXPR_1:%.*]], i32 noundef [[DOTCAPTURE_EXPR_3:%.*]], i32 noundef [[DOTCAPTURE_EXPR_5:%.*]], ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR14]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR4:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR6:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
+// CHECK3-NEXT: store i32 [[DOTCAPTURE_EXPR_1]], ptr [[DOTCAPTURE_EXPR__ADDR2]], align 4
+// CHECK3-NEXT: store i32 [[DOTCAPTURE_EXPR_3]], ptr [[DOTCAPTURE_EXPR__ADDR4]], align 4
+// CHECK3-NEXT: store i32 [[DOTCAPTURE_EXPR_5]], ptr [[DOTCAPTURE_EXPR__ADDR6]], align 4
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR4]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP1]])
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_dynamiciiii_l165.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_dynamiciiii_l165.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_privates_map..55
+// CHECK3-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_10:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_10]], ptr [[TMP4]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_10]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_entry..56
+// CHECK3-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED3_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED4_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED5_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_11:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_11]], ptr [[TMP3]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META140:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META143:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META145:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META147:![0-9]+]])
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META149:![0-9]+]]
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: store ptr @.omp_task_privates_map..55, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META150:![0-9]+]]
+// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_9:%.*]], ptr [[TMP9]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_9]], ptr [[TMP9]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_9]], ptr [[TMP9]], i32 0, i32 3
+// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP19]], i32 [[TMP18]])
+// CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP24:%.*]] = icmp ult i32 [[TMP20]], [[TMP22]]
+// CHECK3-NEXT: [[TMP25:%.*]] = select i1 [[TMP24]], i32 [[TMP20]], i32 [[TMP22]]
+// CHECK3-NEXT: [[TMP26:%.*]] = icmp ult i32 [[TMP21]], [[TMP23]]
+// CHECK3-NEXT: [[TMP27:%.*]] = select i1 [[TMP26]], i32 [[TMP21]], i32 [[TMP23]]
+// CHECK3-NEXT: [[TMP28:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP25]], 0
+// CHECK3-NEXT: [[TMP29:%.*]] = insertvalue [3 x i32] [[TMP28]], i32 [[TMP27]], 1
+// CHECK3-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK3-NEXT: store i32 5, ptr [[TMP30]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP31]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP13]], ptr [[TMP32]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP33]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.53, ptr [[TMP34]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP35]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP36]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP37]], align 8, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP38:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP38]], align 8, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP39]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [[TMP29]], ptr [[TMP40]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP41:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP41]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 [[TMP25]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_dynamiciiii_l165.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK3-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
+// CHECK3-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__54_EXIT:%.*]]
+// CHECK3: omp_offload.failed.i:
+// CHECK3-NEXT: [[TMP44:%.*]] = load i32, ptr [[TMP9]], align 4
+// CHECK3-NEXT: store i32 [[TMP44]], ptr [[DOTCAPTURE_EXPR__CASTED_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP45:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP46:%.*]] = load i32, ptr [[TMP15]], align 4
+// CHECK3-NEXT: store i32 [[TMP46]], ptr [[DOTCAPTURE_EXPR__CASTED3_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP47:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED3_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP48:%.*]] = load i32, ptr [[TMP16]], align 4
+// CHECK3-NEXT: store i32 [[TMP48]], ptr [[DOTCAPTURE_EXPR__CASTED4_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP49:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED4_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP50:%.*]] = load i32, ptr [[TMP17]], align 4
+// CHECK3-NEXT: store i32 [[TMP50]], ptr [[DOTCAPTURE_EXPR__CASTED5_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: [[TMP51:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED5_I]], align 4, !noalias [[META149]]
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_dynamiciiii_l165(i32 [[TMP45]], i32 [[TMP47]], i32 [[TMP49]], i32 [[TMP51]], ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[DOTOMP_OUTLINED__54_EXIT]]
+// CHECK3: .omp_outlined..54.exit:
+// CHECK3-NEXT: ret i32 0
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z30target_parallel_constant_limitii
+// CHECK3-SAME: (i32 noundef [[THREADS_X:%.*]], i32 noundef [[THREADS_Y:%.*]]) #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[THREADS_X_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[THREADS_Y_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED2:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4
+// CHECK3-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_12:%.*]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store i32 [[THREADS_X]], ptr [[THREADS_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[THREADS_Y]], ptr [[THREADS_Y_ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[THREADS_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[THREADS_Y_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK3-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTCAPTURE_EXPR__CASTED2]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED2]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 [[TMP6]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 [[TMP6]], ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr null, ptr [[TMP13]], align 4
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr null, ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_12]], ptr [[AGG_CAPTURED]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK3-NEXT: store i32 [[TMP19]], ptr [[TMP18]], align 4
+// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_12]], ptr [[AGG_CAPTURED]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: store i32 [[TMP21]], ptr [[TMP20]], align 4
+// CHECK3-NEXT: [[TMP22:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 68, i32 8, ptr @.omp_task_entry..61)
+// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_14:%.*]], ptr [[TMP22]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP23]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP25:%.*]] = load ptr, ptr [[TMP24]], align 4
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP25]], ptr align 4 [[AGG_CAPTURED]], i32 8, i1 false)
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_14]], ptr [[TMP22]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_13:%.*]], ptr [[TMP26]], i32 0, i32 0
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP27]], ptr align 4 @.offload_sizes.57, i32 24, i1 false)
+// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_13]], ptr [[TMP26]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP28]], ptr align 4 [[TMP16]], i32 12, i1 false)
+// CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_13]], ptr [[TMP26]], i32 0, i32 2
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP29]], ptr align 4 [[TMP17]], i32 12, i1 false)
+// CHECK3-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP22]])
+// CHECK3-NEXT: [[TMP30:%.*]] = call i32 @.omp_task_entry..61(i32 [[TMP0]], ptr [[TMP22]]) #[[ATTR3]]
+// CHECK3-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP22]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_parallel_constant_limitii_l170
+// CHECK3-SAME: (i32 noundef [[DOTCAPTURE_EXPR_:%.*]], i32 noundef [[DOTCAPTURE_EXPR_1:%.*]], ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR9]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
+// CHECK3-NEXT: store i32 [[DOTCAPTURE_EXPR_1]], ptr [[DOTCAPTURE_EXPR__ADDR2]], align 4
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP1]])
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_parallel_constant_limitii_l170.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_parallel_constant_limitii_l170.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_privates_map..60
+// CHECK3-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_13:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_13]], ptr [[TMP4]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_13]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_entry..61
+// CHECK3-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED3_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_14:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_14]], ptr [[TMP3]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META151:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META154:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META156:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META158:![0-9]+]])
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META160:![0-9]+]]
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: store ptr @.omp_task_privates_map..60, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META161:![0-9]+]]
+// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_12:%.*]], ptr [[TMP9]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP16]], i32 32)
+// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ult i32 32, [[TMP17]]
+// CHECK3-NEXT: [[TMP20:%.*]] = select i1 [[TMP19]], i32 32, i32 [[TMP17]]
+// CHECK3-NEXT: [[TMP21:%.*]] = icmp ult i32 8, [[TMP18]]
+// CHECK3-NEXT: [[TMP22:%.*]] = select i1 [[TMP21]], i32 8, i32 [[TMP18]]
+// CHECK3-NEXT: [[TMP23:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP20]], 0
+// CHECK3-NEXT: [[TMP24:%.*]] = insertvalue [3 x i32] [[TMP23]], i32 [[TMP22]], 1
+// CHECK3-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK3-NEXT: store i32 3, ptr [[TMP25]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP26]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP13]], ptr [[TMP27]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP28]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.58, ptr [[TMP29]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP30]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP31]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP32]], align 8, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP33]], align 8, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP34]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [[TMP24]], ptr [[TMP35]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP36]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP37:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 [[TMP20]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_parallel_constant_limitii_l170.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK3-NEXT: [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
+// CHECK3-NEXT: br i1 [[TMP38]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__59_EXIT:%.*]]
+// CHECK3: omp_offload.failed.i:
+// CHECK3-NEXT: [[TMP39:%.*]] = load i32, ptr [[TMP9]], align 4
+// CHECK3-NEXT: store i32 [[TMP39]], ptr [[DOTCAPTURE_EXPR__CASTED_I]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP40:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED_I]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP41:%.*]] = load i32, ptr [[TMP15]], align 4
+// CHECK3-NEXT: store i32 [[TMP41]], ptr [[DOTCAPTURE_EXPR__CASTED3_I]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: [[TMP42:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED3_I]], align 4, !noalias [[META160]]
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_parallel_constant_limitii_l170(i32 [[TMP40]], i32 [[TMP42]], ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[DOTOMP_OUTLINED__59_EXIT]]
+// CHECK3: .omp_outlined..59.exit:
+// CHECK3-NEXT: ret i32 0
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z29target_parallel_dynamic_limitii
+// CHECK3-SAME: (i32 noundef [[LIMIT_X:%.*]], i32 noundef [[LIMIT_Y:%.*]]) #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[LIMIT_X_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[LIMIT_Y_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED2:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4
+// CHECK3-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_15:%.*]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store i32 [[LIMIT_X]], ptr [[LIMIT_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[LIMIT_Y]], ptr [[LIMIT_Y_ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[LIMIT_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[LIMIT_Y_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK3-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTCAPTURE_EXPR__CASTED2]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED2]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 [[TMP6]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 [[TMP6]], ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr null, ptr [[TMP13]], align 4
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr null, ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_15]], ptr [[AGG_CAPTURED]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK3-NEXT: store i32 [[TMP19]], ptr [[TMP18]], align 4
+// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_15]], ptr [[AGG_CAPTURED]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: store i32 [[TMP21]], ptr [[TMP20]], align 4
+// CHECK3-NEXT: [[TMP22:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 68, i32 8, ptr @.omp_task_entry..66)
+// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_17:%.*]], ptr [[TMP22]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP23]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP25:%.*]] = load ptr, ptr [[TMP24]], align 4
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP25]], ptr align 4 [[AGG_CAPTURED]], i32 8, i1 false)
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_17]], ptr [[TMP22]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_16:%.*]], ptr [[TMP26]], i32 0, i32 0
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP27]], ptr align 4 @.offload_sizes.62, i32 24, i1 false)
+// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_16]], ptr [[TMP26]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP28]], ptr align 4 [[TMP16]], i32 12, i1 false)
+// CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_16]], ptr [[TMP26]], i32 0, i32 2
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP29]], ptr align 4 [[TMP17]], i32 12, i1 false)
+// CHECK3-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP22]])
+// CHECK3-NEXT: [[TMP30:%.*]] = call i32 @.omp_task_entry..66(i32 [[TMP0]], ptr [[TMP22]]) #[[ATTR3]]
+// CHECK3-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP22]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z29target_parallel_dynamic_limitii_l175
+// CHECK3-SAME: (i32 noundef [[DOTCAPTURE_EXPR_:%.*]], i32 noundef [[DOTCAPTURE_EXPR_1:%.*]], ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
+// CHECK3-NEXT: store i32 [[DOTCAPTURE_EXPR_1]], ptr [[DOTCAPTURE_EXPR__ADDR2]], align 4
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z29target_parallel_dynamic_limitii_l175.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z29target_parallel_dynamic_limitii_l175.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_privates_map..65
+// CHECK3-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_16:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_16]], ptr [[TMP4]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_16]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_entry..66
+// CHECK3-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED3_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_17:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_17]], ptr [[TMP3]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META162:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META165:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META167:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META169:![0-9]+]])
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META171:![0-9]+]]
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: store ptr @.omp_task_privates_map..65, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META172:![0-9]+]]
+// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_15:%.*]], ptr [[TMP9]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP17]], i32 [[TMP16]])
+// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP20:%.*]] = icmp ult i32 [[TMP18]], 16
+// CHECK3-NEXT: [[TMP21:%.*]] = select i1 [[TMP20]], i32 [[TMP18]], i32 16
+// CHECK3-NEXT: [[TMP22:%.*]] = icmp ult i32 [[TMP19]], 4
+// CHECK3-NEXT: [[TMP23:%.*]] = select i1 [[TMP22]], i32 [[TMP19]], i32 4
+// CHECK3-NEXT: [[TMP24:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP21]], 0
+// CHECK3-NEXT: [[TMP25:%.*]] = insertvalue [3 x i32] [[TMP24]], i32 [[TMP23]], 1
+// CHECK3-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK3-NEXT: store i32 3, ptr [[TMP26]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP27]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP13]], ptr [[TMP28]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP29]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.63, ptr [[TMP30]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP31]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP32]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP33]], align 8, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP34]], align 8, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP35]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [[TMP25]], ptr [[TMP36]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP37]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 [[TMP21]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z29target_parallel_dynamic_limitii_l175.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK3-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0
+// CHECK3-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__64_EXIT:%.*]]
+// CHECK3: omp_offload.failed.i:
+// CHECK3-NEXT: [[TMP40:%.*]] = load i32, ptr [[TMP9]], align 4
+// CHECK3-NEXT: store i32 [[TMP40]], ptr [[DOTCAPTURE_EXPR__CASTED_I]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED_I]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP42:%.*]] = load i32, ptr [[TMP15]], align 4
+// CHECK3-NEXT: store i32 [[TMP42]], ptr [[DOTCAPTURE_EXPR__CASTED3_I]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: [[TMP43:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED3_I]], align 4, !noalias [[META171]]
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z29target_parallel_dynamic_limitii_l175(i32 [[TMP41]], i32 [[TMP43]], ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[DOTOMP_OUTLINED__64_EXIT]]
+// CHECK3: .omp_outlined..64.exit:
+// CHECK3-NEXT: ret i32 0
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z23target_parallel_minimumv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_18:%.*]], align 1
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 36, i32 1, ptr @.omp_task_entry..71)
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_20:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_20]], ptr [[TMP6]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_19:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP9]], ptr align 4 @.offload_sizes.67, i32 8, i1 false)
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_19]], ptr [[TMP8]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP10]], ptr align 4 [[TMP4]], i32 4, i1 false)
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_19]], ptr [[TMP8]], i32 0, i32 2
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP11]], ptr align 4 [[TMP5]], i32 4, i1 false)
+// CHECK3-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry..71(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK3-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_minimumv_l180
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_minimumv_l180.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_minimumv_l180.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_privates_map..70
+// CHECK3-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_19:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_19]], ptr [[TMP4]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_19]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_entry..71
+// CHECK3-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_20:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_20]], ptr [[TMP3]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META173:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META176:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META178:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META180:![0-9]+]])
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META182:![0-9]+]]
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: store ptr @.omp_task_privates_map..70, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META183:![0-9]+]]
+// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK3-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.68, ptr [[TMP20]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP21]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP22]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP25]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 16, i32 8, i32 0], ptr [[TMP26]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META182]]
+// CHECK3-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_minimumv_l180.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK3-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK3-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__69_EXIT:%.*]]
+// CHECK3: omp_offload.failed.i:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_minimumv_l180(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[DOTOMP_OUTLINED__69_EXIT]]
+// CHECK3: .omp_outlined..69.exit:
+// CHECK3-NEXT: ret i32 0
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z32target_parallel_three_dimensionsv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_21:%.*]], align 1
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 36, i32 1, ptr @.omp_task_entry..76)
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_23:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_23]], ptr [[TMP6]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_22:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP9]], ptr align 4 @.offload_sizes.72, i32 8, i1 false)
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_22]], ptr [[TMP8]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP10]], ptr align 4 [[TMP4]], i32 4, i1 false)
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_22]], ptr [[TMP8]], i32 0, i32 2
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP11]], ptr align 4 [[TMP5]], i32 4, i1 false)
+// CHECK3-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry..76(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK3-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32target_parallel_three_dimensionsv_l185
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32target_parallel_three_dimensionsv_l185.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32target_parallel_three_dimensionsv_l185.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_privates_map..75
+// CHECK3-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_22:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_22]], ptr [[TMP4]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_22]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_entry..76
+// CHECK3-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_23:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_23]], ptr [[TMP3]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META184:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META187:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META189:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META191:![0-9]+]])
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META193:![0-9]+]]
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: store ptr @.omp_task_privates_map..75, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META194:![0-9]+]]
+// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK3-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.73, ptr [[TMP20]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP21]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP22]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP25]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 16, i32 8, i32 2], ptr [[TMP26]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META193]]
+// CHECK3-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32target_parallel_three_dimensionsv_l185.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK3-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK3-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__74_EXIT:%.*]]
+// CHECK3: omp_offload.failed.i:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32target_parallel_three_dimensionsv_l185(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[DOTOMP_OUTLINED__74_EXIT]]
+// CHECK3: .omp_outlined..74.exit:
+// CHECK3-NEXT: ret i32 0
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z24target_parallel_all_onesv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.77, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.78, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 1, i32 0], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_all_onesv_l190.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_all_onesv_l190(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_all_onesv_l190
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR17:[0-9]+]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 1)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_all_onesv_l190.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_all_onesv_l190.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z28nested_target_teams_parallelv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.79, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.80, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 16, i32 4, i32 0], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR11]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195.omp_outlined.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195.omp_outlined.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z43nested_target_teams_distribute_parallel_forv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.81, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.82, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 4, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 16, i32 4, i32 0], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR11]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: store i32 3, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK3: cond.true:
+// CHECK3-NEXT: br label [[COND_END:%.*]]
+// CHECK3: cond.false:
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: br label [[COND_END]]
+// CHECK3: cond.end:
+// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK3: omp.inner.for.cond:
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK3: omp.inner.for.body:
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16)
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201.omp_outlined.omp_outlined, i32 [[TMP7]], i32 [[TMP8]])
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK3: omp.inner.for.inc:
+// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
+// CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK3: omp.loop.exit:
+// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201.omp_outlined.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
+// CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
+// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 3
+// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK3: cond.true:
+// CHECK3-NEXT: br label [[COND_END:%.*]]
+// CHECK3: cond.false:
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: br label [[COND_END]]
+// CHECK3: cond.end:
+// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
+// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK3-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK3: omp.inner.for.cond:
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
+// CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK3: omp.inner.for.body:
+// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
+// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK3: omp.body.continue:
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK3: omp.inner.for.inc:
+// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1
+// CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
+// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK3: omp.inner.for.end:
+// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK3: omp.loop.exit:
+// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z27conditional_target_paralleliii
+// CHECK3-SAME: (i32 noundef [[COND:%.*]], i32 noundef [[THREADS_X:%.*]], i32 noundef [[THREADS_Y:%.*]]) #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[COND_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[THREADS_X_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[THREADS_Y_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED4:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED5:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x ptr], align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: store i32 [[COND]], ptr [[COND_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[THREADS_X]], ptr [[THREADS_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[THREADS_Y]], ptr [[THREADS_Y_ADDR]], align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = load i32, ptr [[COND_ADDR]], align 4
+// CHECK3-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
+// CHECK3-NEXT: [[STOREDV:%.*]] = zext i1 [[TOBOOL]] to i8
+// CHECK3-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[THREADS_X_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[THREADS_Y_ADDR]], align 4
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK3-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP3]], 0
+// CHECK3-NEXT: [[STOREDV3:%.*]] = zext i1 [[LOADEDV]] to i8
+// CHECK3-NEXT: store i8 [[STOREDV3]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1
+// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTCAPTURE_EXPR__CASTED4]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED4]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
+// CHECK3-NEXT: store i32 [[TMP7]], ptr [[DOTCAPTURE_EXPR__CASTED5]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED5]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 [[TMP4]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 [[TMP6]], ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 [[TMP6]], ptr [[TMP13]], align 4
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
+// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
+// CHECK3-NEXT: store i32 [[TMP8]], ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
+// CHECK3-NEXT: store i32 [[TMP8]], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr null, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr null, ptr [[TMP18]], align 4
+// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr null, ptr [[TMP19]], align 4
+// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr null, ptr [[TMP20]], align 4
+// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP23:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK3-NEXT: [[LOADEDV6:%.*]] = icmp ne i8 [[TMP23]], 0
+// CHECK3-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK3-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
+// CHECK3-NEXT: [[TMP26:%.*]] = select i1 [[LOADEDV6]], i32 [[TMP24]], i32 1
+// CHECK3-NEXT: [[TMP27:%.*]] = select i1 [[LOADEDV6]], i32 [[TMP25]], i32 1
+// CHECK3-NEXT: [[TMP28:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP26]], 0
+// CHECK3-NEXT: [[TMP29:%.*]] = insertvalue [3 x i32] [[TMP28]], i32 [[TMP27]], 1
+// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP30]], align 4
+// CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 4, ptr [[TMP31]], align 4
+// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP21]], ptr [[TMP32]], align 4
+// CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP22]], ptr [[TMP33]], align 4
+// CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.83, ptr [[TMP34]], align 4
+// CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.84, ptr [[TMP35]], align 4
+// CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP36]], align 4
+// CHECK3-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP37]], align 4
+// CHECK3-NEXT: [[TMP38:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP38]], align 8
+// CHECK3-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP39]], align 8
+// CHECK3-NEXT: [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP40]], align 4
+// CHECK3-NEXT: [[TMP41:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [[TMP29]], ptr [[TMP41]], align 4
+// CHECK3-NEXT: [[TMP42:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP42]], align 4
+// CHECK3-NEXT: [[TMP43:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 [[TMP26]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP44:%.*]] = icmp ne i32 [[TMP43]], 0
+// CHECK3-NEXT: br i1 [[TMP44]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208(i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP8]], ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208
+// CHECK3-SAME: (i32 noundef [[DOTCAPTURE_EXPR_:%.*]], i32 noundef [[DOTCAPTURE_EXPR_1:%.*]], i32 noundef [[DOTCAPTURE_EXPR_3:%.*]], ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR14]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR2:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR4:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
+// CHECK3-NEXT: store i32 [[DOTCAPTURE_EXPR_1]], ptr [[DOTCAPTURE_EXPR__ADDR2]], align 4
+// CHECK3-NEXT: store i32 [[DOTCAPTURE_EXPR_3]], ptr [[DOTCAPTURE_EXPR__ADDR4]], align 4
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR2]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP1]])
+// CHECK3-NEXT: [[TMP2:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
+// CHECK3-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP2]], 0
+// CHECK3-NEXT: br i1 [[LOADEDV]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
+// CHECK3: omp_if.then:
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208.omp_outlined)
+// CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
+// CHECK3: omp_if.else:
+// CHECK3-NEXT: call void @__kmpc_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTTHREADID_TEMP_]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTBOUND_ZERO_ADDR]], align 4
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208.omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTBOUND_ZERO_ADDR]]) #[[ATTR3]]
+// CHECK3-NEXT: call void @__kmpc_end_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK3-NEXT: br label [[OMP_IF_END]]
+// CHECK3: omp_if.end:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z38conditional_target_parallel_with_limiti
+// CHECK3-SAME: (i32 noundef [[COND:%.*]]) #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[COND_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 4
+// CHECK3-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_24:%.*]], align 1
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store i32 [[COND]], ptr [[COND_ADDR]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[COND_ADDR]], align 4
+// CHECK3-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP1]], 0
+// CHECK3-NEXT: [[STOREDV:%.*]] = zext i1 [[TOBOOL]] to i8
+// CHECK3-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK3-NEXT: [[TMP2:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK3-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP2]], 0
+// CHECK3-NEXT: [[STOREDV1:%.*]] = zext i1 [[LOADEDV]] to i8
+// CHECK3-NEXT: store i8 [[STOREDV1]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1
+// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 [[TMP3]], ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
+// CHECK3-NEXT: store ptr null, ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
+// CHECK3-NEXT: store ptr null, ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
+// CHECK3-NEXT: store ptr null, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON_24]], ptr [[AGG_CAPTURED]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP13:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK3-NEXT: [[LOADEDV2:%.*]] = icmp ne i8 [[TMP13]], 0
+// CHECK3-NEXT: [[STOREDV3:%.*]] = zext i1 [[LOADEDV2]] to i8
+// CHECK3-NEXT: store i8 [[STOREDV3]], ptr [[TMP12]], align 1
+// CHECK3-NEXT: [[TMP14:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 52, i32 1, ptr @.omp_task_entry..89)
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_26:%.*]], ptr [[TMP14]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP15]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP17:%.*]] = load ptr, ptr [[TMP16]], align 4
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[TMP17]], ptr align 1 [[AGG_CAPTURED]], i32 1, i1 false)
+// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_26]], ptr [[TMP14]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_25:%.*]], ptr [[TMP18]], i32 0, i32 0
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP19]], ptr align 4 @.offload_sizes.85, i32 16, i1 false)
+// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_25]], ptr [[TMP18]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP20]], ptr align 4 [[TMP10]], i32 8, i1 false)
+// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_25]], ptr [[TMP18]], i32 0, i32 2
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP21]], ptr align 4 [[TMP11]], i32 8, i1 false)
+// CHECK3-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP14]])
+// CHECK3-NEXT: [[TMP22:%.*]] = call i32 @.omp_task_entry..89(i32 [[TMP0]], ptr [[TMP14]]) #[[ATTR3]]
+// CHECK3-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP14]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213
+// CHECK3-SAME: (i32 noundef [[DOTCAPTURE_EXPR_:%.*]], ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK3-NEXT: [[TMP1:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1
+// CHECK3-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP1]], 0
+// CHECK3-NEXT: br i1 [[LOADEDV]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
+// CHECK3: omp_if.then:
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213.omp_outlined)
+// CHECK3-NEXT: br label [[OMP_IF_END:%.*]]
+// CHECK3: omp_if.else:
+// CHECK3-NEXT: call void @__kmpc_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTTHREADID_TEMP_]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTBOUND_ZERO_ADDR]], align 4
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213.omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTBOUND_ZERO_ADDR]]) #[[ATTR3]]
+// CHECK3-NEXT: call void @__kmpc_end_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK3-NEXT: br label [[OMP_IF_END]]
+// CHECK3: omp_if.end:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_privates_map..88
+// CHECK3-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_25:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_25]], ptr [[TMP4]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_25]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_entry..89
+// CHECK3-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[DOTCAPTURE_EXPR__CASTED_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_26:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_26]], ptr [[TMP3]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META195:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META198:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META200:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META202:![0-9]+]])
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META204:![0-9]+]]
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: store ptr @.omp_task_privates_map..88, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META205:![0-9]+]]
+// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK3-NEXT: [[TMP16:%.*]] = load i8, ptr [[TMP9]], align 1
+// CHECK3-NEXT: [[LOADEDV_I:%.*]] = icmp ne i8 [[TMP16]], 0
+// CHECK3-NEXT: [[TMP17:%.*]] = select i1 [[LOADEDV_I]], i32 16, i32 1
+// CHECK3-NEXT: [[TMP18:%.*]] = select i1 [[LOADEDV_I]], i32 4, i32 1
+// CHECK3-NEXT: [[TMP19:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP17]], 0
+// CHECK3-NEXT: [[TMP20:%.*]] = insertvalue [3 x i32] [[TMP19]], i32 [[TMP18]], 1
+// CHECK3-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK3-NEXT: store i32 2, ptr [[TMP21]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP22]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP13]], ptr [[TMP23]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP24]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.86, ptr [[TMP25]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP26]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP27]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP28]], align 8, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP29]], align 8, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP30]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [[TMP20]], ptr [[TMP31]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP32]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 [[TMP17]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK3-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
+// CHECK3-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__87_EXIT:%.*]]
+// CHECK3: omp_offload.failed.i:
+// CHECK3-NEXT: [[TMP35:%.*]] = load i8, ptr [[TMP9]], align 1
+// CHECK3-NEXT: [[LOADEDV3_I:%.*]] = icmp ne i8 [[TMP35]], 0
+// CHECK3-NEXT: [[STOREDV_I:%.*]] = zext i1 [[LOADEDV3_I]] to i8
+// CHECK3-NEXT: store i8 [[STOREDV_I]], ptr [[DOTCAPTURE_EXPR__CASTED_I]], align 1, !noalias [[META204]]
+// CHECK3-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED_I]], align 4, !noalias [[META204]]
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213(i32 [[TMP36]], ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[DOTOMP_OUTLINED__87_EXIT]]
+// CHECK3: .omp_outlined..87.exit:
+// CHECK3-NEXT: ret i32 0
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z24disabled_target_parallelv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.90, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.91, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24disabled_target_parallelv_l218.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24disabled_target_parallelv_l218(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24disabled_target_parallelv_l218
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK3-NEXT: call void @__kmpc_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTTHREADID_TEMP_]], align 4
+// CHECK3-NEXT: store i32 0, ptr [[DOTBOUND_ZERO_ADDR]], align 4
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24disabled_target_parallelv_l218.omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTBOUND_ZERO_ADDR]]) #[[ATTR3]]
+// CHECK3-NEXT: call void @__kmpc_end_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24disabled_target_parallelv_l218.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z23enabled_target_parallelv
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP0]], align 4
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
+// CHECK3-NEXT: store i32 5, ptr [[TMP5]], align 4
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
+// CHECK3-NEXT: store ptr @.offload_sizes.92, ptr [[TMP9]], align 4
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.93, ptr [[TMP10]], align 4
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
+// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
+// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
+// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8
+// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP15]], align 4
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 16, i32 4, i32 0], ptr [[TMP16]], align 4
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4
+// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23enabled_target_parallelv_l223.region_id, ptr [[KERNEL_ARGS]])
+// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
+// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
+// CHECK3: omp_offload.failed:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23enabled_target_parallelv_l223(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
+// CHECK3: omp_offload.cont:
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23enabled_target_parallelv_l223
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23enabled_target_parallelv_l223.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23enabled_target_parallelv_l223.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z24extra_limit_dimensions_1v
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_27:%.*]], align 1
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 36, i32 1, ptr @.omp_task_entry..98)
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_29:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_29]], ptr [[TMP6]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_28:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP9]], ptr align 4 @.offload_sizes.94, i32 8, i1 false)
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_28]], ptr [[TMP8]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP10]], ptr align 4 [[TMP4]], i32 4, i1 false)
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_28]], ptr [[TMP8]], i32 0, i32 2
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP11]], ptr align 4 [[TMP5]], i32 4, i1 false)
+// CHECK3-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry..98(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK3-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_1v_l230
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_1v_l230.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_1v_l230.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_privates_map..97
+// CHECK3-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_28:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_28]], ptr [[TMP4]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_28]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_entry..98
+// CHECK3-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_29:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_29]], ptr [[TMP3]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META206:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META209:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META211:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META213:![0-9]+]])
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META215:![0-9]+]]
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: store ptr @.omp_task_privates_map..97, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META216:![0-9]+]]
+// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK3-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.95, ptr [[TMP20]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP21]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP22]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP25]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 16, i32 1, i32 0], ptr [[TMP26]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META215]]
+// CHECK3-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_1v_l230.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK3-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK3-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__96_EXIT:%.*]]
+// CHECK3: omp_offload.failed.i:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_1v_l230(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[DOTOMP_OUTLINED__96_EXIT]]
+// CHECK3: .omp_outlined..96.exit:
+// CHECK3-NEXT: ret i32 0
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z24extra_limit_dimensions_2v
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_30:%.*]], align 1
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 36, i32 1, ptr @.omp_task_entry..103)
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_32:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_32]], ptr [[TMP6]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_31:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP9]], ptr align 4 @.offload_sizes.99, i32 8, i1 false)
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_31]], ptr [[TMP8]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP10]], ptr align 4 [[TMP4]], i32 4, i1 false)
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_31]], ptr [[TMP8]], i32 0, i32 2
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP11]], ptr align 4 [[TMP5]], i32 4, i1 false)
+// CHECK3-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry..103(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK3-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_2v_l235
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_2v_l235.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_2v_l235.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_privates_map..102
+// CHECK3-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_31:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_31]], ptr [[TMP4]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_31]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_entry..103
+// CHECK3-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_32:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_32]], ptr [[TMP3]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META217:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META220:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META222:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META224:![0-9]+]])
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META226:![0-9]+]]
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: store ptr @.omp_task_privates_map..102, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META227:![0-9]+]]
+// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK3-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.100, ptr [[TMP20]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP21]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP22]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP25]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 16, i32 1, i32 0], ptr [[TMP26]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META226]]
+// CHECK3-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_2v_l235.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK3-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK3-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__101_EXIT:%.*]]
+// CHECK3: omp_offload.failed.i:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_2v_l235(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[DOTOMP_OUTLINED__101_EXIT]]
+// CHECK3: .omp_outlined..101.exit:
+// CHECK3-NEXT: ret i32 0
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z24extra_limit_dimensions_3v
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_33:%.*]], align 1
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 36, i32 1, ptr @.omp_task_entry..108)
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_35:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_35]], ptr [[TMP6]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_34:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP9]], ptr align 4 @.offload_sizes.104, i32 8, i1 false)
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_34]], ptr [[TMP8]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP10]], ptr align 4 [[TMP4]], i32 4, i1 false)
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_34]], ptr [[TMP8]], i32 0, i32 2
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP11]], ptr align 4 [[TMP5]], i32 4, i1 false)
+// CHECK3-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry..108(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK3-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_3v_l240
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_3v_l240.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_3v_l240.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_privates_map..107
+// CHECK3-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_34:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_34]], ptr [[TMP4]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_34]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_entry..108
+// CHECK3-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_35:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_35]], ptr [[TMP3]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META228:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META231:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META233:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META235:![0-9]+]])
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META237:![0-9]+]]
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: store ptr @.omp_task_privates_map..107, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META238:![0-9]+]]
+// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK3-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.105, ptr [[TMP20]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP21]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP22]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP25]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 16, i32 1, i32 1], ptr [[TMP26]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META237]]
+// CHECK3-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_3v_l240.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK3-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK3-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__106_EXIT:%.*]]
+// CHECK3: omp_offload.failed.i:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_3v_l240(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[DOTOMP_OUTLINED__106_EXIT]]
+// CHECK3: .omp_outlined..106.exit:
+// CHECK3-NEXT: ret i32 0
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z24extra_limit_dimensions_4v
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_36:%.*]], align 1
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 36, i32 1, ptr @.omp_task_entry..113)
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_38:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_38]], ptr [[TMP6]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_37:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP9]], ptr align 4 @.offload_sizes.109, i32 8, i1 false)
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_37]], ptr [[TMP8]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP10]], ptr align 4 [[TMP4]], i32 4, i1 false)
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_37]], ptr [[TMP8]], i32 0, i32 2
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP11]], ptr align 4 [[TMP5]], i32 4, i1 false)
+// CHECK3-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry..113(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK3-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_4v_l245
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_4v_l245.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_4v_l245.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_privates_map..112
+// CHECK3-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_37:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_37]], ptr [[TMP4]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_37]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_entry..113
+// CHECK3-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_38:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_38]], ptr [[TMP3]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META239:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META242:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META244:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META246:![0-9]+]])
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META248:![0-9]+]]
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: store ptr @.omp_task_privates_map..112, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META249:![0-9]+]]
+// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK3-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.110, ptr [[TMP20]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP21]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP22]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP25]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 16, i32 1, i32 1], ptr [[TMP26]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META248]]
+// CHECK3-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_4v_l245.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK3-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK3-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__111_EXIT:%.*]]
+// CHECK3: omp_offload.failed.i:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_4v_l245(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[DOTOMP_OUTLINED__111_EXIT]]
+// CHECK3: .omp_outlined..111.exit:
+// CHECK3-NEXT: ret i32 0
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@_Z24extra_limit_dimensions_5v
+// CHECK3-SAME: () #[[ATTR0]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
+// CHECK3-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_39:%.*]], align 1
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
+// CHECK3-NEXT: store ptr null, ptr [[TMP3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 36, i32 1, ptr @.omp_task_entry..118)
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_41:%.*]], ptr [[TMP6]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_41]], ptr [[TMP6]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_40:%.*]], ptr [[TMP8]], i32 0, i32 0
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP9]], ptr align 4 @.offload_sizes.114, i32 8, i1 false)
+// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_40]], ptr [[TMP8]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP10]], ptr align 4 [[TMP4]], i32 4, i1 false)
+// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_40]], ptr [[TMP8]], i32 0, i32 2
+// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP11]], ptr align 4 [[TMP5]], i32 4, i1 false)
+// CHECK3-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: [[TMP12:%.*]] = call i32 @.omp_task_entry..118(i32 [[TMP0]], ptr [[TMP6]]) #[[ATTR3]]
+// CHECK3-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP6]])
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_5v_l250
+// CHECK3-SAME: (ptr noalias noundef [[DYN_PTR:%.*]]) #[[ATTR15]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK3-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
+// CHECK3-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_5v_l250.omp_outlined)
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_5v_l250.omp_outlined
+// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR10]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_privates_map..117
+// CHECK3-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]], ptr noalias noundef [[TMP2:%.*]], ptr noalias noundef [[TMP3:%.*]]) #[[ATTR6]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR2:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTADDR3:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP2]], ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_40:%.*]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTADDR3]], align 4
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_40]], ptr [[TMP4]], i32 0, i32 1
+// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP8]], align 4
+// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T_40]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTADDR2]], align 4
+// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP10]], align 4
+// CHECK3-NEXT: ret void
+//
+//
+// CHECK3-LABEL: define {{[^@]+}}@.omp_task_entry..118
+// CHECK3-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR7]] {
+// CHECK3-NEXT: entry:
+// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR1_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR2_I:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: [[KERNEL_ARGS_I:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
+// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
+// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
+// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
+// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
+// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_41:%.*]], ptr [[TMP3]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
+// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
+// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
+// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES_41]], ptr [[TMP3]], i32 0, i32 1
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META250:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META253:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META255:![0-9]+]])
+// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META257:![0-9]+]])
+// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META259:![0-9]+]]
+// CHECK3-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: store ptr @.omp_task_privates_map..117, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]]) #[[ATTR3]], !inline_history [[META260:![0-9]+]]
+// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR1_I]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR2_I]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB1]], i32 [[TMP15]], i32 32)
+// CHECK3-NEXT: store i32 5, ptr [[KERNEL_ARGS_I]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 1
+// CHECK3-NEXT: store i32 1, ptr [[TMP16]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 2
+// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP17]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 3
+// CHECK3-NEXT: store ptr [[TMP13]], ptr [[TMP18]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 4
+// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP19]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 5
+// CHECK3-NEXT: store ptr @.offload_maptypes.115, ptr [[TMP20]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 6
+// CHECK3-NEXT: store ptr null, ptr [[TMP21]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 7
+// CHECK3-NEXT: store ptr null, ptr [[TMP22]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 8
+// CHECK3-NEXT: store i64 0, ptr [[TMP23]], align 8, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 9
+// CHECK3-NEXT: store i64 0, ptr [[TMP24]], align 8, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 10
+// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP25]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 11
+// CHECK3-NEXT: store [3 x i32] [i32 16, i32 4, i32 1], ptr [[TMP26]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS_I]], i32 0, i32 12
+// CHECK3-NEXT: store i32 0, ptr [[TMP27]], align 4, !noalias [[META259]]
+// CHECK3-NEXT: [[TMP28:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 16, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_5v_l250.region_id, ptr [[KERNEL_ARGS_I]])
+// CHECK3-NEXT: [[TMP29:%.*]] = icmp ne i32 [[TMP28]], 0
+// CHECK3-NEXT: br i1 [[TMP29]], label [[OMP_OFFLOAD_FAILED_I:%.*]], label [[DOTOMP_OUTLINED__116_EXIT:%.*]]
+// CHECK3: omp_offload.failed.i:
+// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_5v_l250(ptr null) #[[ATTR3]]
+// CHECK3-NEXT: br label [[DOTOMP_OUTLINED__116_EXIT]]
+// CHECK3: .omp_outlined..116.exit:
+// CHECK3-NEXT: ret i32 0
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z25one_dimensional_modifiersv_l47
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1:[0-9]+]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 3, i32 16)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z25one_dimensional_modifiersv_l47.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z25one_dimensional_modifiersv_l47.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_without_clausesv_l52
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR1]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP0]], i32 0
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP1]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27target_teams_num_teams_onlyv_l58
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR3:[0-9]+]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 2, i32 0)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27target_teams_num_teams_onlyv_l58.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27target_teams_num_teams_onlyv_l58.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_num_teams_dynamicii_l63
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR1]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DYN_PTR2:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 8
+// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 1
+// CHECK9-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 8
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 8
+// CHECK9-NEXT: store i32 [[TMP7]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 2
+// CHECK9-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP8]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR2]], align 8
+// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK9-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP9]], i32 0)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_num_teams_dynamicii_l63.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_num_teams_dynamicii_l63.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_thread_limit_onlyv_l68
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR4:[0-9]+]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_thread_limit_onlyv_l68.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_thread_limit_onlyv_l68.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_teams_limitsv_l73
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR0]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 3, i32 16)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_teams_limitsv_l73.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_teams_limitsv_l73.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_thread_limit_onlyv_l78
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR4]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP0]], i32 0
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP1]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_nested_parallelii_l83
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR5:[0-9]+]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[THREADS_X:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[THREADS_Y:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 8
+// CHECK9-NEXT: store i32 [[TMP4]], ptr [[THREADS_X]], align 4
+// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 1
+// CHECK9-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 8
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 8
+// CHECK9-NEXT: store i32 [[TMP7]], ptr [[THREADS_Y]], align 4
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 2
+// CHECK9-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP8]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[THREADS_X]], align 4
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP9]])
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_nested_parallelii_l83.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_nested_parallelii_l83.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6:[0-9]+]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR7:[0-9]+]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89.omp_outlined.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89.omp_outlined.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41nested_teams_limit_overrides_target_limitv_l96
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR4]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41nested_teams_limit_overrides_target_limitv_l96.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41nested_teams_limit_overrides_target_limitv_l96.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_teams_distributev_l102
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR4]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_teams_distributev_l102.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_teams_distributev_l102.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK9: cond.true:
+// CHECK9-NEXT: br label [[COND_END:%.*]]
+// CHECK9: cond.false:
+// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: br label [[COND_END]]
+// CHECK9: cond.end:
+// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK9: omp.inner.for.cond:
+// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK9: omp.inner.for.body:
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK9: omp.body.continue:
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK9: omp.inner.for.inc:
+// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK9-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK9: omp.inner.for.end:
+// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK9: omp.loop.exit:
+// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR7]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK9-NEXT: store i32 3, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK9: cond.true:
+// CHECK9-NEXT: br label [[COND_END:%.*]]
+// CHECK9: cond.false:
+// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: br label [[COND_END]]
+// CHECK9: cond.end:
+// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK9: omp.inner.for.cond:
+// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK9: omp.inner.for.body:
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16)
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
+// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107.omp_outlined.omp_outlined, i64 [[TMP8]], i64 [[TMP10]])
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK9: omp.inner.for.inc:
+// CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
+// CHECK9-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK9: omp.inner.for.end:
+// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK9: omp.loop.exit:
+// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107.omp_outlined.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
+// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
+// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
+// CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
+// CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
+// CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
+// CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
+// CHECK9-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
+// CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 3
+// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK9: cond.true:
+// CHECK9-NEXT: br label [[COND_END:%.*]]
+// CHECK9: cond.false:
+// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: br label [[COND_END]]
+// CHECK9: cond.end:
+// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
+// CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK9: omp.inner.for.cond:
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
+// CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK9: omp.inner.for.body:
+// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
+// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK9: omp.body.continue:
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK9: omp.inner.for.inc:
+// CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
+// CHECK9-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK9: omp.inner.for.end:
+// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK9: omp.loop.exit:
+// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR7]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK9-NEXT: store i32 3, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK9: cond.true:
+// CHECK9-NEXT: br label [[COND_END:%.*]]
+// CHECK9: cond.false:
+// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: br label [[COND_END]]
+// CHECK9: cond.end:
+// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK9: omp.inner.for.cond:
+// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP84:![0-9]+]]
+// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK9: omp.inner.for.body:
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16), !llvm.access.group [[ACC_GRP84]]
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
+// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112.omp_outlined.omp_outlined, i64 [[TMP8]], i64 [[TMP10]]), !llvm.access.group [[ACC_GRP84]]
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK9: omp.inner.for.inc:
+// CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
+// CHECK9-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP85:![0-9]+]]
+// CHECK9: omp.inner.for.end:
+// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK9: omp.loop.exit:
+// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
+// CHECK9-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK9: .omp.final.then:
+// CHECK9-NEXT: store i32 4, ptr [[I]], align 4
+// CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
+// CHECK9: .omp.final.done:
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112.omp_outlined.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
+// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
+// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
+// CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
+// CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
+// CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
+// CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
+// CHECK9-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
+// CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 3
+// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK9: cond.true:
+// CHECK9-NEXT: br label [[COND_END:%.*]]
+// CHECK9: cond.false:
+// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: br label [[COND_END]]
+// CHECK9: cond.end:
+// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
+// CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK9: omp.inner.for.cond:
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP88:![0-9]+]]
+// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP88]]
+// CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
+// CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK9: omp.inner.for.body:
+// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP88]]
+// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
+// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP88]]
+// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK9: omp.body.continue:
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK9: omp.inner.for.inc:
+// CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP88]]
+// CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
+// CHECK9-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP88]]
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP89:![0-9]+]]
+// CHECK9: omp.inner.for.end:
+// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK9: omp.loop.exit:
+// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
+// CHECK9-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK9: .omp.final.then:
+// CHECK9-NEXT: store i32 4, ptr [[I]], align 4
+// CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
+// CHECK9: .omp.final.done:
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_teams_distribute_simdv_l117
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR8:[0-9]+]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 3, i32 32)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_teams_distribute_simdv_l117.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_teams_distribute_simdv_l117.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK9: cond.true:
+// CHECK9-NEXT: br label [[COND_END:%.*]]
+// CHECK9: cond.false:
+// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: br label [[COND_END]]
+// CHECK9: cond.end:
+// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK9: omp.inner.for.cond:
+// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP91:![0-9]+]]
+// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP91]]
+// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK9: omp.inner.for.body:
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP91]]
+// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP91]]
+// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK9: omp.body.continue:
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK9: omp.inner.for.inc:
+// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP91]]
+// CHECK9-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK9-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP91]]
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP92:![0-9]+]]
+// CHECK9: omp.inner.for.end:
+// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK9: omp.loop.exit:
+// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP10:%.*]] = icmp ne i32 [[TMP9]], 0
+// CHECK9-NEXT: br i1 [[TMP10]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK9: .omp.final.then:
+// CHECK9-NEXT: store i32 4, ptr [[I]], align 4
+// CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
+// CHECK9: .omp.final.done:
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR9:[0-9]+]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 3, i32 32)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK9-NEXT: store i32 3, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK9: cond.true:
+// CHECK9-NEXT: br label [[COND_END:%.*]]
+// CHECK9: cond.false:
+// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: br label [[COND_END]]
+// CHECK9: cond.end:
+// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK9: omp.inner.for.cond:
+// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK9: omp.inner.for.body:
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
+// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122.omp_outlined.omp_outlined, i64 [[TMP8]], i64 [[TMP10]])
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK9: omp.inner.for.inc:
+// CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
+// CHECK9-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK9: omp.inner.for.end:
+// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK9: omp.loop.exit:
+// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122.omp_outlined.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
+// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
+// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
+// CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
+// CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
+// CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
+// CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
+// CHECK9-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
+// CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 3
+// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK9: cond.true:
+// CHECK9-NEXT: br label [[COND_END:%.*]]
+// CHECK9: cond.false:
+// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: br label [[COND_END]]
+// CHECK9: cond.end:
+// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
+// CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK9: omp.inner.for.cond:
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
+// CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK9: omp.inner.for.body:
+// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
+// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK9: omp.body.continue:
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK9: omp.inner.for.inc:
+// CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
+// CHECK9-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK9: omp.inner.for.end:
+// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK9: omp.loop.exit:
+// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z33target_parallel_thread_limit_onlyv_l128
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR5]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP0]], i32 0
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP1]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z33target_parallel_thread_limit_onlyv_l128.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z33target_parallel_thread_limit_onlyv_l128.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z31target_parallel_threads_dynamicii_l133
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR10:[0-9]+]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DYN_PTR2:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 8
+// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 1
+// CHECK9-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 8
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 8
+// CHECK9-NEXT: store i32 [[TMP7]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 2
+// CHECK9-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP8]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR2]], align 8
+// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP9]])
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z31target_parallel_threads_dynamicii_l133.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z31target_parallel_threads_dynamicii_l133.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_parallel_one_thread_dimensionv_l138
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11:[0-9]+]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_parallel_one_thread_dimensionv_l138.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_parallel_one_thread_dimensionv_l138.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_threadsv_l143
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR12:[0-9]+]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 8)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_threadsv_l143.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_threadsv_l143.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_parallel_forv_l148
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_parallel_forv_l148.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_parallel_forv_l148.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK9: cond.true:
+// CHECK9-NEXT: br label [[COND_END:%.*]]
+// CHECK9: cond.false:
+// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: br label [[COND_END]]
+// CHECK9: cond.end:
+// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK9: omp.inner.for.cond:
+// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK9: omp.inner.for.body:
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK9: omp.body.continue:
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK9: omp.inner.for.inc:
+// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK9-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK9: omp.inner.for.end:
+// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK9: omp.loop.exit:
+// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_for_simdv_l153
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_for_simdv_l153.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_for_simdv_l153.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK9: cond.true:
+// CHECK9-NEXT: br label [[COND_END:%.*]]
+// CHECK9: cond.false:
+// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: br label [[COND_END]]
+// CHECK9: cond.end:
+// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK9: omp.inner.for.cond:
+// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP94:![0-9]+]]
+// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP94]]
+// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK9: omp.inner.for.body:
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP94]]
+// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP94]]
+// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK9: omp.body.continue:
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK9: omp.inner.for.inc:
+// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP94]]
+// CHECK9-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK9-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP94]]
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP95:![0-9]+]]
+// CHECK9: omp.inner.for.end:
+// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK9: omp.loop.exit:
+// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP10:%.*]] = icmp ne i32 [[TMP9]], 0
+// CHECK9-NEXT: br i1 [[TMP10]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK9: .omp.final.then:
+// CHECK9-NEXT: store i32 4, ptr [[I]], align 4
+// CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
+// CHECK9: .omp.final.done:
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20target_parallel_loopv_l158
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20target_parallel_loopv_l158.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20target_parallel_loopv_l158.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK9: cond.true:
+// CHECK9-NEXT: br label [[COND_END:%.*]]
+// CHECK9: cond.false:
+// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: br label [[COND_END]]
+// CHECK9: cond.end:
+// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK9: omp.inner.for.cond:
+// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK9: omp.inner.for.body:
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK9: omp.body.continue:
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK9: omp.inner.for.inc:
+// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK9-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK9: omp.inner.for.end:
+// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK9: omp.loop.exit:
+// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_dynamiciiii_l165
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR10]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DYN_PTR4:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 8
+// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 1
+// CHECK9-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 8
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 8
+// CHECK9-NEXT: store i32 [[TMP7]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 2
+// CHECK9-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 8
+// CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 8
+// CHECK9-NEXT: store i32 [[TMP10]], ptr [[DOTCAPTURE_EXPR_2]], align 4
+// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 3
+// CHECK9-NEXT: [[TMP12:%.*]] = load ptr, ptr [[TMP11]], align 8
+// CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 8
+// CHECK9-NEXT: store i32 [[TMP13]], ptr [[DOTCAPTURE_EXPR_3]], align 4
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 4
+// CHECK9-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP14]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR4]], align 8
+// CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP15]])
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_dynamiciiii_l165.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_dynamiciiii_l165.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_parallel_constant_limitii_l170
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR5]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DYN_PTR2:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 8
+// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 1
+// CHECK9-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 8
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 8
+// CHECK9-NEXT: store i32 [[TMP7]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 2
+// CHECK9-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP8]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR2]], align 8
+// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP9]])
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_parallel_constant_limitii_l170.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_parallel_constant_limitii_l170.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z29target_parallel_dynamic_limitii_l175
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DYN_PTR2:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 8
+// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 1
+// CHECK9-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 8
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 8
+// CHECK9-NEXT: store i32 [[TMP7]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 2
+// CHECK9-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP8]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR2]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z29target_parallel_dynamic_limitii_l175.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z29target_parallel_dynamic_limitii_l175.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_minimumv_l180
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_minimumv_l180.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_minimumv_l180.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32target_parallel_three_dimensionsv_l185
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32target_parallel_three_dimensionsv_l185.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32target_parallel_three_dimensionsv_l185.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_all_onesv_l190
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR13:[0-9]+]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 1)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_all_onesv_l190.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_all_onesv_l190.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR7]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195.omp_outlined.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195.omp_outlined.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR7]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK9-NEXT: store i32 3, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK9: cond.true:
+// CHECK9-NEXT: br label [[COND_END:%.*]]
+// CHECK9: cond.false:
+// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: br label [[COND_END]]
+// CHECK9: cond.end:
+// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK9: omp.inner.for.cond:
+// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK9: omp.inner.for.body:
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16)
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
+// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201.omp_outlined.omp_outlined, i64 [[TMP8]], i64 [[TMP10]])
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK9: omp.inner.for.inc:
+// CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
+// CHECK9-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK9: omp.inner.for.end:
+// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK9: omp.loop.exit:
+// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201.omp_outlined.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
+// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
+// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
+// CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
+// CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
+// CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
+// CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
+// CHECK9-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
+// CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 3
+// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK9: cond.true:
+// CHECK9-NEXT: br label [[COND_END:%.*]]
+// CHECK9: cond.false:
+// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: br label [[COND_END]]
+// CHECK9: cond.end:
+// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
+// CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK9-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK9: omp.inner.for.cond:
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
+// CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK9: omp.inner.for.body:
+// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
+// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK9: omp.body.continue:
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK9: omp.inner.for.inc:
+// CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
+// CHECK9-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
+// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK9: omp.inner.for.end:
+// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK9: omp.loop.exit:
+// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR10]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
+// CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DYN_PTR3:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: [[TMP4:%.*]] = load i8, ptr [[TMP3]], align 8
+// CHECK9-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP4]], 0
+// CHECK9-NEXT: [[STOREDV:%.*]] = zext i1 [[LOADEDV]] to i8
+// CHECK9-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 1
+// CHECK9-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 8
+// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 8
+// CHECK9-NEXT: store i32 [[TMP7]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 2
+// CHECK9-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 8
+// CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 8
+// CHECK9-NEXT: store i32 [[TMP10]], ptr [[DOTCAPTURE_EXPR_2]], align 4
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 3
+// CHECK9-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP11]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR3]], align 8
+// CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP12]])
+// CHECK9-NEXT: [[TMP13:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK9-NEXT: [[LOADEDV4:%.*]] = icmp ne i8 [[TMP13]], 0
+// CHECK9-NEXT: br i1 [[LOADEDV4]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
+// CHECK9: omp_if.then:
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208.omp_outlined)
+// CHECK9-NEXT: br label [[OMP_IF_END:%.*]]
+// CHECK9: omp_if.else:
+// CHECK9-NEXT: call void @__kmpc_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK9-NEXT: store i32 [[TMP0]], ptr [[DOTTHREADID_TEMP_]], align 4
+// CHECK9-NEXT: store i32 0, ptr [[DOTBOUND_ZERO_ADDR]], align 4
+// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208.omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2:[0-9]+]]
+// CHECK9-NEXT: call void @__kmpc_end_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK9-NEXT: br label [[OMP_IF_END]]
+// CHECK9: omp_if.end:
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: [[TMP4:%.*]] = load i8, ptr [[TMP3]], align 8
+// CHECK9-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP4]], 0
+// CHECK9-NEXT: [[STOREDV:%.*]] = zext i1 [[LOADEDV]] to i8
+// CHECK9-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 1
+// CHECK9-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP5]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK9-NEXT: [[TMP6:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK9-NEXT: [[LOADEDV2:%.*]] = icmp ne i8 [[TMP6]], 0
+// CHECK9-NEXT: br i1 [[LOADEDV2]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
+// CHECK9: omp_if.then:
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213.omp_outlined)
+// CHECK9-NEXT: br label [[OMP_IF_END:%.*]]
+// CHECK9: omp_if.else:
+// CHECK9-NEXT: call void @__kmpc_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK9-NEXT: store i32 [[TMP0]], ptr [[DOTTHREADID_TEMP_]], align 4
+// CHECK9-NEXT: store i32 0, ptr [[DOTBOUND_ZERO_ADDR]], align 4
+// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213.omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2]]
+// CHECK9-NEXT: call void @__kmpc_end_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK9-NEXT: br label [[OMP_IF_END]]
+// CHECK9: omp_if.end:
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24disabled_target_parallelv_l218
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK9-NEXT: call void @__kmpc_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK9-NEXT: store i32 [[TMP0]], ptr [[DOTTHREADID_TEMP_]], align 4
+// CHECK9-NEXT: store i32 0, ptr [[DOTBOUND_ZERO_ADDR]], align 4
+// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24disabled_target_parallelv_l218.omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2]]
+// CHECK9-NEXT: call void @__kmpc_end_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24disabled_target_parallelv_l218.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23enabled_target_parallelv_l223
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23enabled_target_parallelv_l223.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23enabled_target_parallelv_l223.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_1v_l230
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_1v_l230.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_1v_l230.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_2v_l235
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_2v_l235.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_2v_l235.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_3v_l240
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_3v_l240.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_3v_l240.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_4v_l245
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_4v_l245.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_4v_l245.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_5v_l250
+// CHECK9-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK9-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 8
+// CHECK9-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 8
+// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 8
+// CHECK9-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_5v_l250.omp_outlined)
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_5v_l250.omp_outlined
+// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK9-NEXT: entry:
+// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
+// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
+// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
+// CHECK9-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z25one_dimensional_modifiersv_l47
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1:[0-9]+]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 3, i32 16)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z25one_dimensional_modifiersv_l47.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z25one_dimensional_modifiersv_l47.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_without_clausesv_l52
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR1]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP0]], i32 0
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP1]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27target_teams_num_teams_onlyv_l58
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR3:[0-9]+]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 2, i32 0)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27target_teams_num_teams_onlyv_l58.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27target_teams_num_teams_onlyv_l58.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_num_teams_dynamicii_l63
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR1]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DYN_PTR2:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
+// CHECK11-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 1
+// CHECK11-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 4
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
+// CHECK11-NEXT: store i32 [[TMP7]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 2
+// CHECK11-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP8]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR2]], align 4
+// CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP9]], i32 0)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_num_teams_dynamicii_l63.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_num_teams_dynamicii_l63.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_thread_limit_onlyv_l68
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR4:[0-9]+]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_thread_limit_onlyv_l68.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_teams_thread_limit_onlyv_l68.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_teams_limitsv_l73
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR0]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 3, i32 16)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_teams_limitsv_l73.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_teams_limitsv_l73.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_thread_limit_onlyv_l78
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR4]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP0]], i32 0
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP1]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_nested_parallelii_l83
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR5:[0-9]+]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[THREADS_X:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[THREADS_Y:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
+// CHECK11-NEXT: store i32 [[TMP4]], ptr [[THREADS_X]], align 4
+// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 1
+// CHECK11-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 4
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
+// CHECK11-NEXT: store i32 [[TMP7]], ptr [[THREADS_Y]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 2
+// CHECK11-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP8]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[THREADS_X]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP9]])
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_nested_parallelii_l83.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z22target_nested_parallelii_l83.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6:[0-9]+]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR7:[0-9]+]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89.omp_outlined.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_nested_teams_parallelv_l89.omp_outlined.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41nested_teams_limit_overrides_target_limitv_l96
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR4]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41nested_teams_limit_overrides_target_limitv_l96.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41nested_teams_limit_overrides_target_limitv_l96.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_teams_distributev_l102
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR4]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_teams_distributev_l102.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_teams_distributev_l102.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK11: cond.true:
+// CHECK11-NEXT: br label [[COND_END:%.*]]
+// CHECK11: cond.false:
+// CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: br label [[COND_END]]
+// CHECK11: cond.end:
+// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK11: omp.inner.for.cond:
+// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK11: omp.inner.for.body:
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK11-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK11: omp.body.continue:
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK11: omp.inner.for.inc:
+// CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK11-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK11: omp.inner.for.end:
+// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK11: omp.loop.exit:
+// CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR7]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK11-NEXT: store i32 3, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK11: cond.true:
+// CHECK11-NEXT: br label [[COND_END:%.*]]
+// CHECK11: cond.false:
+// CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: br label [[COND_END]]
+// CHECK11: cond.end:
+// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK11-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK11: omp.inner.for.cond:
+// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK11: omp.inner.for.body:
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16)
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107.omp_outlined.omp_outlined, i32 [[TMP7]], i32 [[TMP8]])
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK11: omp.inner.for.inc:
+// CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
+// CHECK11-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK11: omp.inner.for.end:
+// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK11: omp.loop.exit:
+// CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_teams_distribute_parallel_forv_l107.omp_outlined.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
+// CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
+// CHECK11-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
+// CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 3
+// CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK11: cond.true:
+// CHECK11-NEXT: br label [[COND_END:%.*]]
+// CHECK11: cond.false:
+// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: br label [[COND_END]]
+// CHECK11: cond.end:
+// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
+// CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK11: omp.inner.for.cond:
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
+// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK11: omp.inner.for.body:
+// CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
+// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK11-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK11: omp.body.continue:
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK11: omp.inner.for.inc:
+// CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1
+// CHECK11-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK11: omp.inner.for.end:
+// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK11: omp.loop.exit:
+// CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR7]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK11-NEXT: store i32 3, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK11: cond.true:
+// CHECK11-NEXT: br label [[COND_END:%.*]]
+// CHECK11: cond.false:
+// CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: br label [[COND_END]]
+// CHECK11: cond.end:
+// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK11-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK11: omp.inner.for.cond:
+// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP84:![0-9]+]]
+// CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK11: omp.inner.for.body:
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16), !llvm.access.group [[ACC_GRP84]]
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112.omp_outlined.omp_outlined, i32 [[TMP7]], i32 [[TMP8]]), !llvm.access.group [[ACC_GRP84]]
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK11: omp.inner.for.inc:
+// CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
+// CHECK11-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP84]]
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP85:![0-9]+]]
+// CHECK11: omp.inner.for.end:
+// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK11: omp.loop.exit:
+// CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
+// CHECK11-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK11: .omp.final.then:
+// CHECK11-NEXT: store i32 4, ptr [[I]], align 4
+// CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
+// CHECK11: .omp.final.done:
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z41target_teams_distribute_parallel_for_simdv_l112.omp_outlined.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
+// CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
+// CHECK11-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
+// CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 3
+// CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK11: cond.true:
+// CHECK11-NEXT: br label [[COND_END:%.*]]
+// CHECK11: cond.false:
+// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: br label [[COND_END]]
+// CHECK11: cond.end:
+// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
+// CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK11: omp.inner.for.cond:
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP88:![0-9]+]]
+// CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP88]]
+// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
+// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK11: omp.inner.for.body:
+// CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP88]]
+// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
+// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK11-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP88]]
+// CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK11: omp.body.continue:
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK11: omp.inner.for.inc:
+// CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP88]]
+// CHECK11-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1
+// CHECK11-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP88]]
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP89:![0-9]+]]
+// CHECK11: omp.inner.for.end:
+// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK11: omp.loop.exit:
+// CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
+// CHECK11-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK11: .omp.final.then:
+// CHECK11-NEXT: store i32 4, ptr [[I]], align 4
+// CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
+// CHECK11: .omp.final.done:
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_teams_distribute_simdv_l117
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR8:[0-9]+]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 3, i32 32)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_teams_distribute_simdv_l117.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28target_teams_distribute_simdv_l117.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK11: cond.true:
+// CHECK11-NEXT: br label [[COND_END:%.*]]
+// CHECK11: cond.false:
+// CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: br label [[COND_END]]
+// CHECK11: cond.end:
+// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK11: omp.inner.for.cond:
+// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP91:![0-9]+]]
+// CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP91]]
+// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK11: omp.inner.for.body:
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP91]]
+// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK11-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP91]]
+// CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK11: omp.body.continue:
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK11: omp.inner.for.inc:
+// CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP91]]
+// CHECK11-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK11-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP91]]
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP92:![0-9]+]]
+// CHECK11: omp.inner.for.end:
+// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK11: omp.loop.exit:
+// CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP10:%.*]] = icmp ne i32 [[TMP9]], 0
+// CHECK11-NEXT: br i1 [[TMP10]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK11: .omp.final.then:
+// CHECK11-NEXT: store i32 4, ptr [[I]], align 4
+// CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
+// CHECK11: .omp.final.done:
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR9:[0-9]+]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 3, i32 32)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK11-NEXT: store i32 3, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK11: cond.true:
+// CHECK11-NEXT: br label [[COND_END:%.*]]
+// CHECK11: cond.false:
+// CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: br label [[COND_END]]
+// CHECK11: cond.end:
+// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK11-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK11: omp.inner.for.cond:
+// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK11: omp.inner.for.body:
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122.omp_outlined.omp_outlined, i32 [[TMP7]], i32 [[TMP8]])
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK11: omp.inner.for.inc:
+// CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
+// CHECK11-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK11: omp.inner.for.end:
+// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK11: omp.loop.exit:
+// CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17target_teams_loopv_l122.omp_outlined.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
+// CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
+// CHECK11-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
+// CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 3
+// CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK11: cond.true:
+// CHECK11-NEXT: br label [[COND_END:%.*]]
+// CHECK11: cond.false:
+// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: br label [[COND_END]]
+// CHECK11: cond.end:
+// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
+// CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK11: omp.inner.for.cond:
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
+// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK11: omp.inner.for.body:
+// CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
+// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK11-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK11: omp.body.continue:
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK11: omp.inner.for.inc:
+// CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1
+// CHECK11-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK11: omp.inner.for.end:
+// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK11: omp.loop.exit:
+// CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z33target_parallel_thread_limit_onlyv_l128
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR5]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP0]], i32 0
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP1]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z33target_parallel_thread_limit_onlyv_l128.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z33target_parallel_thread_limit_onlyv_l128.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z31target_parallel_threads_dynamicii_l133
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR10:[0-9]+]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DYN_PTR2:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
+// CHECK11-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 1
+// CHECK11-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 4
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
+// CHECK11-NEXT: store i32 [[TMP7]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 2
+// CHECK11-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP8]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR2]], align 4
+// CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP9]])
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z31target_parallel_threads_dynamicii_l133.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z31target_parallel_threads_dynamicii_l133.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_parallel_one_thread_dimensionv_l138
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11:[0-9]+]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_parallel_one_thread_dimensionv_l138.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z36target_parallel_one_thread_dimensionv_l138.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_threadsv_l143
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR12:[0-9]+]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 8)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_threadsv_l143.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_threadsv_l143.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_parallel_forv_l148
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_parallel_forv_l148.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19target_parallel_forv_l148.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK11: cond.true:
+// CHECK11-NEXT: br label [[COND_END:%.*]]
+// CHECK11: cond.false:
+// CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: br label [[COND_END]]
+// CHECK11: cond.end:
+// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK11: omp.inner.for.cond:
+// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK11: omp.inner.for.body:
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK11-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK11: omp.body.continue:
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK11: omp.inner.for.inc:
+// CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK11-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK11: omp.inner.for.end:
+// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK11: omp.loop.exit:
+// CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_for_simdv_l153
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_for_simdv_l153.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_for_simdv_l153.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK11: cond.true:
+// CHECK11-NEXT: br label [[COND_END:%.*]]
+// CHECK11: cond.false:
+// CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: br label [[COND_END]]
+// CHECK11: cond.end:
+// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK11: omp.inner.for.cond:
+// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP94:![0-9]+]]
+// CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP94]]
+// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK11: omp.inner.for.body:
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP94]]
+// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK11-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP94]]
+// CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK11: omp.body.continue:
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK11: omp.inner.for.inc:
+// CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP94]]
+// CHECK11-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK11-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP94]]
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP95:![0-9]+]]
+// CHECK11: omp.inner.for.end:
+// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK11: omp.loop.exit:
+// CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP10:%.*]] = icmp ne i32 [[TMP9]], 0
+// CHECK11-NEXT: br i1 [[TMP10]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
+// CHECK11: .omp.final.then:
+// CHECK11-NEXT: store i32 4, ptr [[I]], align 4
+// CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]]
+// CHECK11: .omp.final.done:
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20target_parallel_loopv_l158
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20target_parallel_loopv_l158.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z20target_parallel_loopv_l158.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK11: cond.true:
+// CHECK11-NEXT: br label [[COND_END:%.*]]
+// CHECK11: cond.false:
+// CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: br label [[COND_END]]
+// CHECK11: cond.end:
+// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK11: omp.inner.for.cond:
+// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK11: omp.inner.for.body:
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
+// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK11-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK11: omp.body.continue:
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK11: omp.inner.for.inc:
+// CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 1
+// CHECK11-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK11: omp.inner.for.end:
+// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK11: omp.loop.exit:
+// CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP1]])
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_dynamiciiii_l165
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR10]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DYN_PTR4:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
+// CHECK11-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 1
+// CHECK11-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 4
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
+// CHECK11-NEXT: store i32 [[TMP7]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 2
+// CHECK11-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 4
+// CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
+// CHECK11-NEXT: store i32 [[TMP10]], ptr [[DOTCAPTURE_EXPR_2]], align 4
+// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 3
+// CHECK11-NEXT: [[TMP12:%.*]] = load ptr, ptr [[TMP11]], align 4
+// CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
+// CHECK11-NEXT: store i32 [[TMP13]], ptr [[DOTCAPTURE_EXPR_3]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 4
+// CHECK11-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP14]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR4]], align 4
+// CHECK11-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP15]])
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_dynamiciiii_l165.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_dynamiciiii_l165.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_parallel_constant_limitii_l170
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR5]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DYN_PTR2:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
+// CHECK11-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 1
+// CHECK11-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 4
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
+// CHECK11-NEXT: store i32 [[TMP7]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 2
+// CHECK11-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP8]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR2]], align 4
+// CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP9]])
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_parallel_constant_limitii_l170.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z30target_parallel_constant_limitii_l170.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z29target_parallel_dynamic_limitii_l175
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DYN_PTR2:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
+// CHECK11-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_]], align 4
+// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 1
+// CHECK11-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 4
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
+// CHECK11-NEXT: store i32 [[TMP7]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 2
+// CHECK11-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP8]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR2]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z29target_parallel_dynamic_limitii_l175.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z29target_parallel_dynamic_limitii_l175.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_minimumv_l180
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_minimumv_l180.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23target_parallel_minimumv_l180.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32target_parallel_three_dimensionsv_l185
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32target_parallel_three_dimensionsv_l185.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32target_parallel_three_dimensionsv_l185.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_all_onesv_l190
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR13:[0-9]+]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 1)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_all_onesv_l190.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24target_parallel_all_onesv_l190.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR7]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195.omp_outlined.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z28nested_target_teams_parallelv_l195.omp_outlined.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR7]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB1]], i32 [[TMP0]], i32 0, i32 32)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK11-NEXT: store i32 3, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+// CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 3
+// CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK11: cond.true:
+// CHECK11-NEXT: br label [[COND_END:%.*]]
+// CHECK11: cond.false:
+// CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: br label [[COND_END]]
+// CHECK11: cond.end:
+// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
+// CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK11-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK11: omp.inner.for.cond:
+// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
+// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK11: omp.inner.for.body:
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP1]], i32 16)
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
+// CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201.omp_outlined.omp_outlined, i32 [[TMP7]], i32 [[TMP8]])
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK11: omp.inner.for.inc:
+// CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
+// CHECK11-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK11: omp.inner.for.end:
+// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK11: omp.loop.exit:
+// CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z43nested_target_teams_distribute_parallel_forv_l201.omp_outlined.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
+// CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 3, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
+// CHECK11-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
+// CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
+// CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 3
+// CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
+// CHECK11: cond.true:
+// CHECK11-NEXT: br label [[COND_END:%.*]]
+// CHECK11: cond.false:
+// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: br label [[COND_END]]
+// CHECK11: cond.end:
+// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 3, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
+// CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
+// CHECK11-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
+// CHECK11: omp.inner.for.cond:
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
+// CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
+// CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
+// CHECK11: omp.inner.for.body:
+// CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
+// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
+// CHECK11-NEXT: store i32 [[ADD]], ptr [[I]], align 4
+// CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
+// CHECK11: omp.body.continue:
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
+// CHECK11: omp.inner.for.inc:
+// CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1
+// CHECK11-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
+// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]]
+// CHECK11: omp.inner.for.end:
+// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
+// CHECK11: omp.loop.exit:
+// CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP3]])
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR10]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
+// CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DYN_PTR3:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: [[TMP4:%.*]] = load i8, ptr [[TMP3]], align 4
+// CHECK11-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP4]], 0
+// CHECK11-NEXT: [[STOREDV:%.*]] = zext i1 [[LOADEDV]] to i8
+// CHECK11-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 1
+// CHECK11-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 4
+// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
+// CHECK11-NEXT: store i32 [[TMP7]], ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 2
+// CHECK11-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 4
+// CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
+// CHECK11-NEXT: store i32 [[TMP10]], ptr [[DOTCAPTURE_EXPR_2]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 3
+// CHECK11-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP11]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR3]], align 4
+// CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 [[TMP12]])
+// CHECK11-NEXT: [[TMP13:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK11-NEXT: [[LOADEDV4:%.*]] = icmp ne i8 [[TMP13]], 0
+// CHECK11-NEXT: br i1 [[LOADEDV4]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
+// CHECK11: omp_if.then:
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208.omp_outlined)
+// CHECK11-NEXT: br label [[OMP_IF_END:%.*]]
+// CHECK11: omp_if.else:
+// CHECK11-NEXT: call void @__kmpc_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK11-NEXT: store i32 [[TMP0]], ptr [[DOTTHREADID_TEMP_]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTBOUND_ZERO_ADDR]], align 4
+// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208.omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2:[0-9]+]]
+// CHECK11-NEXT: call void @__kmpc_end_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK11-NEXT: br label [[OMP_IF_END]]
+// CHECK11: omp_if.end:
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z27conditional_target_paralleliii_l208.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: [[TMP4:%.*]] = load i8, ptr [[TMP3]], align 4
+// CHECK11-NEXT: [[LOADEDV:%.*]] = icmp ne i8 [[TMP4]], 0
+// CHECK11-NEXT: [[STOREDV:%.*]] = zext i1 [[LOADEDV]] to i8
+// CHECK11-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 1
+// CHECK11-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP5]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK11-NEXT: [[TMP6:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1
+// CHECK11-NEXT: [[LOADEDV2:%.*]] = icmp ne i8 [[TMP6]], 0
+// CHECK11-NEXT: br i1 [[LOADEDV2]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
+// CHECK11: omp_if.then:
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213.omp_outlined)
+// CHECK11-NEXT: br label [[OMP_IF_END:%.*]]
+// CHECK11: omp_if.else:
+// CHECK11-NEXT: call void @__kmpc_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK11-NEXT: store i32 [[TMP0]], ptr [[DOTTHREADID_TEMP_]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTBOUND_ZERO_ADDR]], align 4
+// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213.omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2]]
+// CHECK11-NEXT: call void @__kmpc_end_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK11-NEXT: br label [[OMP_IF_END]]
+// CHECK11: omp_if.end:
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z38conditional_target_parallel_with_limiti_l213.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24disabled_target_parallelv_l218
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[DOTBOUND_ZERO_ADDR:%.*]] = alloca i32, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK11-NEXT: call void @__kmpc_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK11-NEXT: store i32 [[TMP0]], ptr [[DOTTHREADID_TEMP_]], align 4
+// CHECK11-NEXT: store i32 0, ptr [[DOTBOUND_ZERO_ADDR]], align 4
+// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24disabled_target_parallelv_l218.omp_outlined(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTBOUND_ZERO_ADDR]]) #[[ATTR2]]
+// CHECK11-NEXT: call void @__kmpc_end_serialized_parallel(ptr @[[GLOB1]], i32 [[TMP0]])
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24disabled_target_parallelv_l218.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23enabled_target_parallelv_l223
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23enabled_target_parallelv_l223.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z23enabled_target_parallelv_l223.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_1v_l230
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_1v_l230.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_1v_l230.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_2v_l235
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_2v_l235.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_2v_l235.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_3v_l240
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_3v_l240.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_3v_l240.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_4v_l245
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_4v_l245.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_4v_l245.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_5v_l250
+// CHECK11-SAME: (ptr noalias noundef [[__CONTEXT:%.*]]) #[[ATTR11]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DYN_PTR1:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
+// CHECK11-NEXT: store ptr [[__CONTEXT]], ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__CONTEXT_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DYN_PTR_ADDR]], align 4
+// CHECK11-NEXT: [[DYN_PTR:%.*]] = load ptr, ptr [[TMP2]], align 4
+// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR1]], align 4
+// CHECK11-NEXT: call void @__kmpc_push_num_threads(ptr @[[GLOB1]], i32 [[TMP0]], i32 16)
+// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_5v_l250.omp_outlined)
+// CHECK11-NEXT: ret void
+//
+//
+// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z24extra_limit_dimensions_5v_l250.omp_outlined
+// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR6]] {
+// CHECK11-NEXT: entry:
+// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
+// CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
+// CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
+// CHECK11-NEXT: ret void
+//
More information about the cfe-commits
mailing list