/build/source/llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp

Bug Summary

File:	build/source/llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp
Warning:	line 1941, column 20 Called C++ object pointer is null
Annotated Source Code

Press '?' to see keyboard shortcuts
Show analyzer invocation
clang -cc1 -cc1 -triple x86_64-pc-linux-gnu -analyze -disable-free -clear-ast-before-backend -disable-llvm-verifier -discard-value-names -main-file-name OMPIRBuilder.cpp -analyzer-checker=core -analyzer-checker=apiModeling -analyzer-checker=unix -analyzer-checker=deadcode -analyzer-checker=cplusplus -analyzer-checker=security.insecureAPI.UncheckedReturn -analyzer-checker=security.insecureAPI.getpw -analyzer-checker=security.insecureAPI.gets -analyzer-checker=security.insecureAPI.mktemp -analyzer-checker=security.insecureAPI.mkstemp -analyzer-checker=security.insecureAPI.vfork -analyzer-checker=nullability.NullPassedToNonnull -analyzer-checker=nullability.NullReturnedFromNonnull -analyzer-output plist -w -setup-static-analyzer -analyzer-config-compatibility-mode=true -mrelocation-model pic -pic-level 2 -mframe-pointer=none -fmath-errno -ffp-contract=on -fno-rounding-math -mconstructor-aliases -funwind-tables=2 -target-cpu x86-64 -tune-cpu generic -debugger-tuning=gdb -ffunction-sections -fdata-sections -fcoverage-compilation-dir=/build/source/build-llvm/tools/clang/stage2-bins -resource-dir /usr/lib/llvm-17/lib/clang/17 -D _DEBUG -D _GLIBCXX_ASSERTIONS -D _GNU_SOURCE -D _LIBCPP_ENABLE_ASSERTIONS -D __STDC_CONSTANT_MACROS -D __STDC_FORMAT_MACROS -D __STDC_LIMIT_MACROS -I lib/Frontend/OpenMP -I /build/source/llvm/lib/Frontend/OpenMP -I include -I /build/source/llvm/include -D _FORTIFY_SOURCE=2 -D NDEBUG -U NDEBUG -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/10/../../../../include/c++/10 -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/10/../../../../include/x86_64-linux-gnu/c++/10 -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/10/../../../../include/c++/10/backward -internal-isystem /usr/lib/llvm-17/lib/clang/17/include -internal-isystem /usr/local/include -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/10/../../../../x86_64-linux-gnu/include -internal-externc-isystem /usr/include/x86_64-linux-gnu -internal-externc-isystem /include -internal-externc-isystem /usr/include -fmacro-prefix-map=/build/source/build-llvm/tools/clang/stage2-bins=build-llvm/tools/clang/stage2-bins -fmacro-prefix-map=/build/source/= -fcoverage-prefix-map=/build/source/build-llvm/tools/clang/stage2-bins=build-llvm/tools/clang/stage2-bins -fcoverage-prefix-map=/build/source/= -source-date-epoch 1683717183 -O2 -Wno-unused-command-line-argument -Wno-unused-parameter -Wwrite-strings -Wno-missing-field-initializers -Wno-long-long -Wno-maybe-uninitialized -Wno-class-memaccess -Wno-redundant-move -Wno-pessimizing-move -Wno-noexcept-type -Wno-comment -Wno-misleading-indentation -std=c++17 -fdeprecated-macro -fdebug-compilation-dir=/build/source/build-llvm/tools/clang/stage2-bins -fdebug-prefix-map=/build/source/build-llvm/tools/clang/stage2-bins=build-llvm/tools/clang/stage2-bins -fdebug-prefix-map=/build/source/= -ferror-limit 19 -fvisibility-inlines-hidden -stack-protector 2 -fgnuc-version=4.2.1 -fcolor-diagnostics -vectorize-loops -vectorize-slp -analyzer-output=html -analyzer-config stable-report-filename=true -faddrsig -D__GCC_HAVE_DWARF2_CFI_ASM=1 -o /tmp/scan-build-2023-05-10-133810-16478-1 -x c++ /build/source/llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp
1//===- OpenMPIRBuilder.cpp - Builder for LLVM-IR for OpenMP directives ----===//
2//
3// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4// See https://llvm.org/LICENSE.txt for license information.
5// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6//
7//===----------------------------------------------------------------------===//
8/// \file
9///
10/// This file implements the OpenMPIRBuilder class, which is used as a
11/// convenient way to create LLVM instructions for OpenMP directives.
12///
13//===----------------------------------------------------------------------===//

15#include "llvm/Frontend/OpenMP/OMPIRBuilder.h"
16#include "llvm/ADT/SmallSet.h"
17#include "llvm/ADT/StringRef.h"
18#include "llvm/Analysis/AssumptionCache.h"
19#include "llvm/Analysis/CodeMetrics.h"
20#include "llvm/Analysis/LoopInfo.h"
21#include "llvm/Analysis/OptimizationRemarkEmitter.h"
22#include "llvm/Analysis/ScalarEvolution.h"
23#include "llvm/Analysis/TargetLibraryInfo.h"
24#include "llvm/IR/CFG.h"
25#include "llvm/IR/Constants.h"
26#include "llvm/IR/DebugInfoMetadata.h"
27#include "llvm/IR/DerivedTypes.h"
28#include "llvm/IR/GlobalVariable.h"
29#include "llvm/IR/IRBuilder.h"
30#include "llvm/IR/MDBuilder.h"
31#include "llvm/IR/PassManager.h"
32#include "llvm/IR/Value.h"
33#include "llvm/MC/TargetRegistry.h"
34#include "llvm/Support/CommandLine.h"
35#include "llvm/Target/TargetMachine.h"
36#include "llvm/Target/TargetOptions.h"
37#include "llvm/Transforms/Utils/BasicBlockUtils.h"
38#include "llvm/Transforms/Utils/Cloning.h"
39#include "llvm/Transforms/Utils/CodeExtractor.h"
40#include "llvm/Transforms/Utils/LoopPeel.h"
41#include "llvm/Transforms/Utils/UnrollLoop.h"

43#include <cstdint>
44#include <optional>

46#define DEBUG_TYPE"openmp-ir-builder" "openmp-ir-builder"

48using namespace llvm;
49using namespace omp;

51static cl::opt<bool>
  OptimisticAttributes("openmp-ir-builder-optimistic-attributes", cl::Hidden,
                       cl::desc("Use optimistic attributes describing "
                                "'as-if' properties of runtime calls."),
                       cl::init(false));

57static cl::opt<double> UnrollThresholdFactor(
  "openmp-ir-builder-unroll-threshold-factor", cl::Hidden,
  cl::desc("Factor for the unroll threshold to account for code "
           "simplifications still taking place"),
  cl::init(1.5));

63#ifndef NDEBUG
64/// Return whether IP1 and IP2 are ambiguous, i.e. that inserting instructions
65/// at position IP1 may change the meaning of IP2 or vice-versa. This is because
66/// an InsertPoint stores the instruction before something is inserted. For
67/// instance, if both point to the same instruction, two IRBuilders alternating
68/// creating instruction will cause the instructions to be interleaved.
69static bool isConflictIP(IRBuilder<>::InsertPoint IP1,
                       IRBuilder<>::InsertPoint IP2) {
if (!IP1.isSet() || !IP2.isSet())
  return false;
return IP1.getBlock() == IP2.getBlock() && IP1.getPoint() == IP2.getPoint();
74}

76static bool isValidWorkshareLoopScheduleType(OMPScheduleType SchedType) {
// Valid ordered/unordered and base algorithm combinations.
switch (SchedType & ~OMPScheduleType::MonotonicityMask) {
case OMPScheduleType::UnorderedStaticChunked:
case OMPScheduleType::UnorderedStatic:
case OMPScheduleType::UnorderedDynamicChunked:
case OMPScheduleType::UnorderedGuidedChunked:
case OMPScheduleType::UnorderedRuntime:
case OMPScheduleType::UnorderedAuto:
case OMPScheduleType::UnorderedTrapezoidal:
case OMPScheduleType::UnorderedGreedy:
case OMPScheduleType::UnorderedBalanced:
case OMPScheduleType::UnorderedGuidedIterativeChunked:
case OMPScheduleType::UnorderedGuidedAnalyticalChunked:
case OMPScheduleType::UnorderedSteal:
case OMPScheduleType::UnorderedStaticBalancedChunked:
case OMPScheduleType::UnorderedGuidedSimd:
case OMPScheduleType::UnorderedRuntimeSimd:
case OMPScheduleType::OrderedStaticChunked:
case OMPScheduleType::OrderedStatic:
case OMPScheduleType::OrderedDynamicChunked:
case OMPScheduleType::OrderedGuidedChunked:
case OMPScheduleType::OrderedRuntime:
case OMPScheduleType::OrderedAuto:
case OMPScheduleType::OrderdTrapezoidal:
case OMPScheduleType::NomergeUnorderedStaticChunked:
case OMPScheduleType::NomergeUnorderedStatic:
case OMPScheduleType::NomergeUnorderedDynamicChunked:
case OMPScheduleType::NomergeUnorderedGuidedChunked:
case OMPScheduleType::NomergeUnorderedRuntime:
case OMPScheduleType::NomergeUnorderedAuto:
case OMPScheduleType::NomergeUnorderedTrapezoidal:
case OMPScheduleType::NomergeUnorderedGreedy:
case OMPScheduleType::NomergeUnorderedBalanced:
case OMPScheduleType::NomergeUnorderedGuidedIterativeChunked:
case OMPScheduleType::NomergeUnorderedGuidedAnalyticalChunked:
case OMPScheduleType::NomergeUnorderedSteal:
case OMPScheduleType::NomergeOrderedStaticChunked:
case OMPScheduleType::NomergeOrderedStatic:
case OMPScheduleType::NomergeOrderedDynamicChunked:
case OMPScheduleType::NomergeOrderedGuidedChunked:
case OMPScheduleType::NomergeOrderedRuntime:
case OMPScheduleType::NomergeOrderedAuto:
case OMPScheduleType::NomergeOrderedTrapezoidal:
  break;
default:
  return false;
}

// Must not set both monotonicity modifiers at the same time.
OMPScheduleType MonotonicityFlags =
    SchedType & OMPScheduleType::MonotonicityMask;
if (MonotonicityFlags == OMPScheduleType::MonotonicityMask)
  return false;

return true;
132}
133#endif

135/// Determine which scheduling algorithm to use, determined from schedule clause
136/// arguments.
137static OMPScheduleType
138getOpenMPBaseScheduleType(llvm::omp::ScheduleKind ClauseKind, bool HasChunks,
                        bool HasSimdModifier) {
// Currently, the default schedule it static.
switch (ClauseKind) {
case OMP_SCHEDULE_Default:
case OMP_SCHEDULE_Static:
  return HasChunks ? OMPScheduleType::BaseStaticChunked
                   : OMPScheduleType::BaseStatic;
case OMP_SCHEDULE_Dynamic:
  return OMPScheduleType::BaseDynamicChunked;
case OMP_SCHEDULE_Guided:
  return HasSimdModifier ? OMPScheduleType::BaseGuidedSimd
                         : OMPScheduleType::BaseGuidedChunked;
case OMP_SCHEDULE_Auto:
  return llvm::omp::OMPScheduleType::BaseAuto;
case OMP_SCHEDULE_Runtime:
  return HasSimdModifier ? OMPScheduleType::BaseRuntimeSimd
                         : OMPScheduleType::BaseRuntime;
}
llvm_unreachable("unhandled schedule clause argument")::llvm::llvm_unreachable_internal("unhandled schedule clause argument"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 157);
158}

160/// Adds ordering modifier flags to schedule type.
161static OMPScheduleType
162getOpenMPOrderingScheduleType(OMPScheduleType BaseScheduleType,
                            bool HasOrderedClause) {
assert((BaseScheduleType & OMPScheduleType::ModifierMask) ==(static_cast <bool> ((BaseScheduleType & OMPScheduleType
::ModifierMask) == OMPScheduleType::None && "Must not have ordering nor monotonicity flags already set"
) ? void (0) : __assert_fail ("(BaseScheduleType & OMPScheduleType::ModifierMask) == OMPScheduleType::None && \"Must not have ordering nor monotonicity flags already set\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 166, __extension__
 __PRETTY_FUNCTION__))
           OMPScheduleType::None &&(static_cast <bool> ((BaseScheduleType & OMPScheduleType
::ModifierMask) == OMPScheduleType::None && "Must not have ordering nor monotonicity flags already set"
) ? void (0) : __assert_fail ("(BaseScheduleType & OMPScheduleType::ModifierMask) == OMPScheduleType::None && \"Must not have ordering nor monotonicity flags already set\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 166, __extension__
 __PRETTY_FUNCTION__))
       "Must not have ordering nor monotonicity flags already set")(static_cast <bool> ((BaseScheduleType & OMPScheduleType
::ModifierMask) == OMPScheduleType::None && "Must not have ordering nor monotonicity flags already set"
) ? void (0) : __assert_fail ("(BaseScheduleType & OMPScheduleType::ModifierMask) == OMPScheduleType::None && \"Must not have ordering nor monotonicity flags already set\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 166, __extension__
 __PRETTY_FUNCTION__));

OMPScheduleType OrderingModifier = HasOrderedClause
                                       ? OMPScheduleType::ModifierOrdered
                                       : OMPScheduleType::ModifierUnordered;
OMPScheduleType OrderingScheduleType = BaseScheduleType | OrderingModifier;

// Unsupported combinations
if (OrderingScheduleType ==
    (OMPScheduleType::BaseGuidedSimd | OMPScheduleType::ModifierOrdered))
  return OMPScheduleType::OrderedGuidedChunked;
else if (OrderingScheduleType == (OMPScheduleType::BaseRuntimeSimd |
                                  OMPScheduleType::ModifierOrdered))
  return OMPScheduleType::OrderedRuntime;

return OrderingScheduleType;
182}

184/// Adds monotonicity modifier flags to schedule type.
185static OMPScheduleType
186getOpenMPMonotonicityScheduleType(OMPScheduleType ScheduleType,
                                bool HasSimdModifier, bool HasMonotonic,
                                bool HasNonmonotonic, bool HasOrderedClause) {
assert((ScheduleType & OMPScheduleType::MonotonicityMask) ==(static_cast <bool> ((ScheduleType & OMPScheduleType
::MonotonicityMask) == OMPScheduleType::None && "Must not have monotonicity flags already set"
) ? void (0) : __assert_fail ("(ScheduleType & OMPScheduleType::MonotonicityMask) == OMPScheduleType::None && \"Must not have monotonicity flags already set\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 191, __extension__
 __PRETTY_FUNCTION__))
           OMPScheduleType::None &&(static_cast <bool> ((ScheduleType & OMPScheduleType
::MonotonicityMask) == OMPScheduleType::None && "Must not have monotonicity flags already set"
) ? void (0) : __assert_fail ("(ScheduleType & OMPScheduleType::MonotonicityMask) == OMPScheduleType::None && \"Must not have monotonicity flags already set\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 191, __extension__
 __PRETTY_FUNCTION__))
       "Must not have monotonicity flags already set")(static_cast <bool> ((ScheduleType & OMPScheduleType
::MonotonicityMask) == OMPScheduleType::None && "Must not have monotonicity flags already set"
) ? void (0) : __assert_fail ("(ScheduleType & OMPScheduleType::MonotonicityMask) == OMPScheduleType::None && \"Must not have monotonicity flags already set\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 191, __extension__
 __PRETTY_FUNCTION__));
assert((!HasMonotonic || !HasNonmonotonic) &&(static_cast <bool> ((!HasMonotonic || !HasNonmonotonic
) && "Monotonic and Nonmonotonic are contradicting each other"
) ? void (0) : __assert_fail ("(!HasMonotonic || !HasNonmonotonic) && \"Monotonic and Nonmonotonic are contradicting each other\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 193, __extension__
 __PRETTY_FUNCTION__))
       "Monotonic and Nonmonotonic are contradicting each other")(static_cast <bool> ((!HasMonotonic || !HasNonmonotonic
) && "Monotonic and Nonmonotonic are contradicting each other"
) ? void (0) : __assert_fail ("(!HasMonotonic || !HasNonmonotonic) && \"Monotonic and Nonmonotonic are contradicting each other\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 193, __extension__
 __PRETTY_FUNCTION__));

if (HasMonotonic) {
  return ScheduleType | OMPScheduleType::ModifierMonotonic;
} else if (HasNonmonotonic) {
  return ScheduleType | OMPScheduleType::ModifierNonmonotonic;
} else {
  // OpenMP 5.1, 2.11.4 Worksharing-Loop Construct, Description.
  // If the static schedule kind is specified or if the ordered clause is
  // specified, and if the nonmonotonic modifier is not specified, the
  // effect is as if the monotonic modifier is specified. Otherwise, unless
  // the monotonic modifier is specified, the effect is as if the
  // nonmonotonic modifier is specified.
  OMPScheduleType BaseScheduleType =
      ScheduleType & ~OMPScheduleType::ModifierMask;
  if ((BaseScheduleType == OMPScheduleType::BaseStatic) ||
      (BaseScheduleType == OMPScheduleType::BaseStaticChunked) ||
      HasOrderedClause) {
    // The monotonic is used by default in openmp runtime library, so no need
    // to set it.
    return ScheduleType;
  } else {
    return ScheduleType | OMPScheduleType::ModifierNonmonotonic;
  }
}
218}

220/// Determine the schedule type using schedule and ordering clause arguments.
221static OMPScheduleType
222computeOpenMPScheduleType(ScheduleKind ClauseKind, bool HasChunks,
                        bool HasSimdModifier, bool HasMonotonicModifier,
                        bool HasNonmonotonicModifier, bool HasOrderedClause) {
OMPScheduleType BaseSchedule =
    getOpenMPBaseScheduleType(ClauseKind, HasChunks, HasSimdModifier);
OMPScheduleType OrderedSchedule =
    getOpenMPOrderingScheduleType(BaseSchedule, HasOrderedClause);
OMPScheduleType Result = getOpenMPMonotonicityScheduleType(
    OrderedSchedule, HasSimdModifier, HasMonotonicModifier,
    HasNonmonotonicModifier, HasOrderedClause);

assert(isValidWorkshareLoopScheduleType(Result))(static_cast <bool> (isValidWorkshareLoopScheduleType(Result
)) ? void (0) : __assert_fail ("isValidWorkshareLoopScheduleType(Result)"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 233, __extension__
 __PRETTY_FUNCTION__));
return Result;
235}

237/// Make \p Source branch to \p Target.
238///
239/// Handles two situations:
240/// * \p Source already has an unconditional branch.
241/// * \p Source is a degenerate block (no terminator because the BB is
242///             the current head of the IR construction).
243static void redirectTo(BasicBlock *Source, BasicBlock *Target, DebugLoc DL) {
if (Instruction *Term = Source->getTerminator()) {
  auto *Br = cast<BranchInst>(Term);
  assert(!Br->isConditional() &&(static_cast <bool> (!Br->isConditional() &&
 "BB's terminator must be an unconditional branch (or degenerate)"
) ? void (0) : __assert_fail ("!Br->isConditional() && \"BB's terminator must be an unconditional branch (or degenerate)\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 247, __extension__
 __PRETTY_FUNCTION__))
         "BB's terminator must be an unconditional branch (or degenerate)")(static_cast <bool> (!Br->isConditional() &&
 "BB's terminator must be an unconditional branch (or degenerate)"
) ? void (0) : __assert_fail ("!Br->isConditional() && \"BB's terminator must be an unconditional branch (or degenerate)\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 247, __extension__
 __PRETTY_FUNCTION__));
  BasicBlock *Succ = Br->getSuccessor(0);
  Succ->removePredecessor(Source, /*KeepOneInputPHIs=*/true);
  Br->setSuccessor(0, Target);
  return;
}

auto *NewBr = BranchInst::Create(Target, Source);
NewBr->setDebugLoc(DL);
256}

258void llvm::spliceBB(IRBuilderBase::InsertPoint IP, BasicBlock *New,
                  bool CreateBranch) {
assert(New->getFirstInsertionPt() == New->begin() &&(static_cast <bool> (New->getFirstInsertionPt() == New
->begin() && "Target BB must not have PHI nodes") ?
 void (0) : __assert_fail ("New->getFirstInsertionPt() == New->begin() && \"Target BB must not have PHI nodes\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 261, __extension__
 __PRETTY_FUNCTION__))
       "Target BB must not have PHI nodes")(static_cast <bool> (New->getFirstInsertionPt() == New
->begin() && "Target BB must not have PHI nodes") ?
 void (0) : __assert_fail ("New->getFirstInsertionPt() == New->begin() && \"Target BB must not have PHI nodes\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 261, __extension__
 __PRETTY_FUNCTION__));

// Move instructions to new block.
BasicBlock *Old = IP.getBlock();
New->splice(New->begin(), Old, IP.getPoint(), Old->end());

if (CreateBranch)
  BranchInst::Create(New, Old);
269}

271void llvm::spliceBB(IRBuilder<> &Builder, BasicBlock *New, bool CreateBranch) {
DebugLoc DebugLoc = Builder.getCurrentDebugLocation();
BasicBlock *Old = Builder.GetInsertBlock();

spliceBB(Builder.saveIP(), New, CreateBranch);
if (CreateBranch)
  Builder.SetInsertPoint(Old->getTerminator());
else
  Builder.SetInsertPoint(Old);

// SetInsertPoint also updates the Builder's debug location, but we want to
// keep the one the Builder was configured to use.
Builder.SetCurrentDebugLocation(DebugLoc);
284}

286BasicBlock *llvm::splitBB(IRBuilderBase::InsertPoint IP, bool CreateBranch,
                        llvm::Twine Name) {
BasicBlock *Old = IP.getBlock();
BasicBlock *New = BasicBlock::Create(
    Old->getContext(), Name.isTriviallyEmpty() ? Old->getName() : Name,
    Old->getParent(), Old->getNextNode());
spliceBB(IP, New, CreateBranch);
New->replaceSuccessorsPhiUsesWith(Old, New);
return New;
295}

297BasicBlock *llvm::splitBB(IRBuilderBase &Builder, bool CreateBranch,
                        llvm::Twine Name) {
DebugLoc DebugLoc = Builder.getCurrentDebugLocation();
BasicBlock *New = splitBB(Builder.saveIP(), CreateBranch, Name);
if (CreateBranch)
  Builder.SetInsertPoint(Builder.GetInsertBlock()->getTerminator());
else
  Builder.SetInsertPoint(Builder.GetInsertBlock());
// SetInsertPoint also updates the Builder's debug location, but we want to
// keep the one the Builder was configured to use.
Builder.SetCurrentDebugLocation(DebugLoc);
return New;
309}

311BasicBlock *llvm::splitBB(IRBuilder<> &Builder, bool CreateBranch,
                        llvm::Twine Name) {
DebugLoc DebugLoc = Builder.getCurrentDebugLocation();
BasicBlock *New = splitBB(Builder.saveIP(), CreateBranch, Name);
if (CreateBranch)
  Builder.SetInsertPoint(Builder.GetInsertBlock()->getTerminator());
else
  Builder.SetInsertPoint(Builder.GetInsertBlock());
// SetInsertPoint also updates the Builder's debug location, but we want to
// keep the one the Builder was configured to use.
Builder.SetCurrentDebugLocation(DebugLoc);
return New;
323}

325BasicBlock *llvm::splitBBWithSuffix(IRBuilderBase &Builder, bool CreateBranch,
                                  llvm::Twine Suffix) {
BasicBlock *Old = Builder.GetInsertBlock();
return splitBB(Builder, CreateBranch, Old->getName() + Suffix);
329}

331void OpenMPIRBuilder::addAttributes(omp::RuntimeFunction FnID, Function &Fn) {
LLVMContext &Ctx = Fn.getContext();
Triple T(M.getTargetTriple());

// Get the function's current attributes.
auto Attrs = Fn.getAttributes();
auto FnAttrs = Attrs.getFnAttrs();
auto RetAttrs = Attrs.getRetAttrs();
SmallVector<AttributeSet, 4> ArgAttrs;
for (size_t ArgNo = 0; ArgNo < Fn.arg_size(); ++ArgNo)
  ArgAttrs.emplace_back(Attrs.getParamAttrs(ArgNo));

// Add AS to FnAS while taking special care with integer extensions.
auto addAttrSet = [&](AttributeSet &FnAS, const AttributeSet &AS,
                      bool Param = true) -> void {
  bool HasSignExt = AS.hasAttribute(Attribute::SExt);
  bool HasZeroExt = AS.hasAttribute(Attribute::ZExt);
  if (HasSignExt || HasZeroExt) {
    assert(AS.getNumAttributes() == 1 &&(static_cast <bool> (AS.getNumAttributes() == 1 &&
 "Currently not handling extension attr combined with others."
) ? void (0) : __assert_fail ("AS.getNumAttributes() == 1 && \"Currently not handling extension attr combined with others.\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 350, __extension__
 __PRETTY_FUNCTION__))
           "Currently not handling extension attr combined with others.")(static_cast <bool> (AS.getNumAttributes() == 1 &&
 "Currently not handling extension attr combined with others."
) ? void (0) : __assert_fail ("AS.getNumAttributes() == 1 && \"Currently not handling extension attr combined with others.\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 350, __extension__
 __PRETTY_FUNCTION__));
    if (Param) {
      if (auto AK = TargetLibraryInfo::getExtAttrForI32Param(T, HasSignExt))
        FnAS = FnAS.addAttribute(Ctx, AK);
    } else
      if (auto AK = TargetLibraryInfo::getExtAttrForI32Return(T, HasSignExt))
        FnAS = FnAS.addAttribute(Ctx, AK);
  } else {
    FnAS = FnAS.addAttributes(Ctx, AS);
  }
};

362#define OMP_ATTRS_SET(VarName, AttrSet) AttributeSet VarName = AttrSet;
363#include "llvm/Frontend/OpenMP/OMPKinds.def"

// Add attributes to the function declaration.
switch (FnID) {
367#define OMP_RTL_ATTRS(Enum, FnAttrSet, RetAttrSet, ArgAttrSets)                \
case Enum:                                                                   \
  FnAttrs = FnAttrs.addAttributes(Ctx, FnAttrSet);                           \
  addAttrSet(RetAttrs, RetAttrSet, /*Param*/false);                          \
  for (size_t ArgNo = 0; ArgNo < ArgAttrSets.size(); ++ArgNo)                \
    addAttrSet(ArgAttrs[ArgNo], ArgAttrSets[ArgNo]);                         \
  Fn.setAttributes(AttributeList::get(Ctx, FnAttrs, RetAttrs, ArgAttrs));    \
  break;
375#include "llvm/Frontend/OpenMP/OMPKinds.def"
default:
  // Attributes are optional.
  break;
}
380}

382FunctionCallee
383OpenMPIRBuilder::getOrCreateRuntimeFunction(Module &M, RuntimeFunction FnID) {
FunctionType *FnTy = nullptr;
Function *Fn = nullptr;

// Try to find the declation in the module first.
switch (FnID) {
389#define OMP_RTL(Enum, Str, IsVarArg, ReturnType, ...)                          \
case Enum:                                                                   \
  FnTy = FunctionType::get(ReturnType, ArrayRef<Type *>{__VA_ARGS__},        \
                           IsVarArg);                                        \
  Fn = M.getFunction(Str);                                                   \
  break;
395#include "llvm/Frontend/OpenMP/OMPKinds.def"
}

if (!Fn) {
  // Create a new declaration if we need one.
  switch (FnID) {
401#define OMP_RTL(Enum, Str, ...)                                                \
case Enum:                                                                   \
  Fn = Function::Create(FnTy, GlobalValue::ExternalLinkage, Str, M);         \
  break;
405#include "llvm/Frontend/OpenMP/OMPKinds.def"
  }

  // Add information if the runtime function takes a callback function
  if (FnID == OMPRTL___kmpc_fork_call || FnID == OMPRTL___kmpc_fork_teams) {
    if (!Fn->hasMetadata(LLVMContext::MD_callback)) {
      LLVMContext &Ctx = Fn->getContext();
      MDBuilder MDB(Ctx);
      // Annotate the callback behavior of the runtime function:
      //  - The callback callee is argument number 2 (microtask).
      //  - The first two arguments of the callback callee are unknown (-1).
      //  - All variadic arguments to the runtime function are passed to the
      //    callback callee.
      Fn->addMetadata(
          LLVMContext::MD_callback,
          *MDNode::get(Ctx, {MDB.createCallbackEncoding(
                                2, {-1, -1}, /* VarArgsArePassed */ true)}));
    }
  }

  LLVM_DEBUG(dbgs() << "Created OpenMP runtime function " << Fn->getName()do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Created OpenMP runtime function "
 << Fn->getName() << " with type " << *Fn
->getFunctionType() << "\n"; } } while (false)
                    << " with type " << *Fn->getFunctionType() << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Created OpenMP runtime function "
 << Fn->getName() << " with type " << *Fn
->getFunctionType() << "\n"; } } while (false);
  addAttributes(FnID, *Fn);

} else {
  LLVM_DEBUG(dbgs() << "Found OpenMP runtime function " << Fn->getName()do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Found OpenMP runtime function "
 << Fn->getName() << " with type " << *Fn
->getFunctionType() << "\n"; } } while (false)
                    << " with type " << *Fn->getFunctionType() << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Found OpenMP runtime function "
 << Fn->getName() << " with type " << *Fn
->getFunctionType() << "\n"; } } while (false);
}

assert(Fn && "Failed to create OpenMP runtime function")(static_cast <bool> (Fn && "Failed to create OpenMP runtime function"
) ? void (0) : __assert_fail ("Fn && \"Failed to create OpenMP runtime function\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 434, __extension__
 __PRETTY_FUNCTION__));

// Cast the function to the expected type if necessary
Constant *C = ConstantExpr::getBitCast(Fn, FnTy->getPointerTo());
return {FnTy, C};
439}

441Function *OpenMPIRBuilder::getOrCreateRuntimeFunctionPtr(RuntimeFunction FnID) {
FunctionCallee RTLFn = getOrCreateRuntimeFunction(M, FnID);
auto *Fn = dyn_cast<llvm::Function>(RTLFn.getCallee());
assert(Fn && "Failed to create OpenMP runtime function pointer")(static_cast <bool> (Fn && "Failed to create OpenMP runtime function pointer"
) ? void (0) : __assert_fail ("Fn && \"Failed to create OpenMP runtime function pointer\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 444, __extension__
 __PRETTY_FUNCTION__));
return Fn;
446}

448void OpenMPIRBuilder::initialize() { initializeTypes(M); }

450void OpenMPIRBuilder::finalize(Function *Fn) {
SmallPtrSet<BasicBlock *, 32> ParallelRegionBlockSet;
SmallVector<BasicBlock *, 32> Blocks;
SmallVector<OutlineInfo, 16> DeferredOutlines;
for (OutlineInfo &OI : OutlineInfos) {
  // Skip functions that have not finalized yet; may happen with nested
  // function generation.
  if (Fn && OI.getFunction() != Fn) {
    DeferredOutlines.push_back(OI);
    continue;
  }

  ParallelRegionBlockSet.clear();
  Blocks.clear();
  OI.collectBlocks(ParallelRegionBlockSet, Blocks);

  Function *OuterFn = OI.getFunction();
  CodeExtractorAnalysisCache CEAC(*OuterFn);
  CodeExtractor Extractor(Blocks, /* DominatorTree */ nullptr,
                          /* AggregateArgs */ true,
                          /* BlockFrequencyInfo */ nullptr,
                          /* BranchProbabilityInfo */ nullptr,
                          /* AssumptionCache */ nullptr,
                          /* AllowVarArgs */ true,
                          /* AllowAlloca */ true,
                          /* AllocaBlock*/ OI.OuterAllocaBB,
                          /* Suffix */ ".omp_par");

  LLVM_DEBUG(dbgs() << "Before     outlining: " << *OuterFn << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Before     outlining: "
 << *OuterFn << "\n"; } } while (false);
  LLVM_DEBUG(dbgs() << "Entry " << OI.EntryBB->getName()do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Entry " << OI
.EntryBB->getName() << " Exit: " << OI.ExitBB->
getName() << "\n"; } } while (false)
                    << " Exit: " << OI.ExitBB->getName() << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Entry " << OI
.EntryBB->getName() << " Exit: " << OI.ExitBB->
getName() << "\n"; } } while (false);
  assert(Extractor.isEligible() &&(static_cast <bool> (Extractor.isEligible() && "Expected OpenMP outlining to be possible!"
) ? void (0) : __assert_fail ("Extractor.isEligible() && \"Expected OpenMP outlining to be possible!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 482, __extension__
 __PRETTY_FUNCTION__))
         "Expected OpenMP outlining to be possible!")(static_cast <bool> (Extractor.isEligible() && "Expected OpenMP outlining to be possible!"
) ? void (0) : __assert_fail ("Extractor.isEligible() && \"Expected OpenMP outlining to be possible!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 482, __extension__
 __PRETTY_FUNCTION__));

  for (auto *V : OI.ExcludeArgsFromAggregate)
    Extractor.excludeArgFromAggregate(V);

  Function *OutlinedFn = Extractor.extractCodeRegion(CEAC);

  LLVM_DEBUG(dbgs() << "After      outlining: " << *OuterFn << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "After      outlining: "
 << *OuterFn << "\n"; } } while (false);
  LLVM_DEBUG(dbgs() << "   Outlined function: " << *OutlinedFn << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "   Outlined function: "
 << *OutlinedFn << "\n"; } } while (false);
  assert(OutlinedFn->getReturnType()->isVoidTy() &&(static_cast <bool> (OutlinedFn->getReturnType()->
isVoidTy() && "OpenMP outlined functions should not return a value!"
) ? void (0) : __assert_fail ("OutlinedFn->getReturnType()->isVoidTy() && \"OpenMP outlined functions should not return a value!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 492, __extension__
 __PRETTY_FUNCTION__))
         "OpenMP outlined functions should not return a value!")(static_cast <bool> (OutlinedFn->getReturnType()->
isVoidTy() && "OpenMP outlined functions should not return a value!"
) ? void (0) : __assert_fail ("OutlinedFn->getReturnType()->isVoidTy() && \"OpenMP outlined functions should not return a value!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 492, __extension__
 __PRETTY_FUNCTION__));

  // For compability with the clang CG we move the outlined function after the
  // one with the parallel region.
  OutlinedFn->removeFromParent();
  M.getFunctionList().insertAfter(OuterFn->getIterator(), OutlinedFn);

  // Remove the artificial entry introduced by the extractor right away, we
  // made our own entry block after all.
  {
    BasicBlock &ArtificialEntry = OutlinedFn->getEntryBlock();
    assert(ArtificialEntry.getUniqueSuccessor() == OI.EntryBB)(static_cast <bool> (ArtificialEntry.getUniqueSuccessor
() == OI.EntryBB) ? void (0) : __assert_fail ("ArtificialEntry.getUniqueSuccessor() == OI.EntryBB"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 503, __extension__
 __PRETTY_FUNCTION__));
    assert(OI.EntryBB->getUniquePredecessor() == &ArtificialEntry)(static_cast <bool> (OI.EntryBB->getUniquePredecessor
() == &ArtificialEntry) ? void (0) : __assert_fail ("OI.EntryBB->getUniquePredecessor() == &ArtificialEntry"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 504, __extension__
 __PRETTY_FUNCTION__));
    // Move instructions from the to-be-deleted ArtificialEntry to the entry
    // basic block of the parallel region. CodeExtractor generates
    // instructions to unwrap the aggregate argument and may sink
    // allocas/bitcasts for values that are solely used in the outlined region
    // and do not escape.
    assert(!ArtificialEntry.empty() &&(static_cast <bool> (!ArtificialEntry.empty() &&
 "Expected instructions to add in the outlined region entry")
 ? void (0) : __assert_fail ("!ArtificialEntry.empty() && \"Expected instructions to add in the outlined region entry\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 511, __extension__
 __PRETTY_FUNCTION__))
           "Expected instructions to add in the outlined region entry")(static_cast <bool> (!ArtificialEntry.empty() &&
 "Expected instructions to add in the outlined region entry")
 ? void (0) : __assert_fail ("!ArtificialEntry.empty() && \"Expected instructions to add in the outlined region entry\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 511, __extension__
 __PRETTY_FUNCTION__));
    for (BasicBlock::reverse_iterator It = ArtificialEntry.rbegin(),
                                      End = ArtificialEntry.rend();
         It != End;) {
      Instruction &I = *It;
      It++;

      if (I.isTerminator())
        continue;

      I.moveBefore(*OI.EntryBB, OI.EntryBB->getFirstInsertionPt());
    }

    OI.EntryBB->moveBefore(&ArtificialEntry);
    ArtificialEntry.eraseFromParent();
  }
  assert(&OutlinedFn->getEntryBlock() == OI.EntryBB)(static_cast <bool> (&OutlinedFn->getEntryBlock(
) == OI.EntryBB) ? void (0) : __assert_fail ("&OutlinedFn->getEntryBlock() == OI.EntryBB"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 527, __extension__
 __PRETTY_FUNCTION__));
  assert(OutlinedFn && OutlinedFn->getNumUses() == 1)(static_cast <bool> (OutlinedFn && OutlinedFn->
getNumUses() == 1) ? void (0) : __assert_fail ("OutlinedFn && OutlinedFn->getNumUses() == 1"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 528, __extension__
 __PRETTY_FUNCTION__));

  // Run a user callback, e.g. to add attributes.
  if (OI.PostOutlineCB)
    OI.PostOutlineCB(*OutlinedFn);
}

// Remove work items that have been completed.
OutlineInfos = std::move(DeferredOutlines);
537}

539OpenMPIRBuilder::~OpenMPIRBuilder() {
assert(OutlineInfos.empty() && "There must be no outstanding outlinings")(static_cast <bool> (OutlineInfos.empty() && "There must be no outstanding outlinings"
) ? void (0) : __assert_fail ("OutlineInfos.empty() && \"There must be no outstanding outlinings\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 540, __extension__
 __PRETTY_FUNCTION__));
541}

543GlobalValue *OpenMPIRBuilder::createGlobalFlag(unsigned Value, StringRef Name) {
IntegerType *I32Ty = Type::getInt32Ty(M.getContext());
auto *GV =
    new GlobalVariable(M, I32Ty,
                       /* isConstant = */ true, GlobalValue::WeakODRLinkage,
                       ConstantInt::get(I32Ty, Value), Name);
GV->setVisibility(GlobalValue::HiddenVisibility);

return GV;
552}

554Constant *OpenMPIRBuilder::getOrCreateIdent(Constant *SrcLocStr,
                                          uint32_t SrcLocStrSize,
                                          IdentFlag LocFlags,
                                          unsigned Reserve2Flags) {
// Enable "C-mode".
LocFlags |= OMP_IDENT_FLAG_KMPC;

Constant *&Ident =
    IdentMap[{SrcLocStr, uint64_t(LocFlags) << 31 | Reserve2Flags}];
if (!Ident) {
  Constant *I32Null = ConstantInt::getNullValue(Int32);
  Constant *IdentData[] = {I32Null,
                           ConstantInt::get(Int32, uint32_t(LocFlags)),
                           ConstantInt::get(Int32, Reserve2Flags),
                           ConstantInt::get(Int32, SrcLocStrSize), SrcLocStr};
  Constant *Initializer =
      ConstantStruct::get(OpenMPIRBuilder::Ident, IdentData);

  // Look for existing encoding of the location + flags, not needed but
  // minimizes the difference to the existing solution while we transition.
  for (GlobalVariable &GV : M.globals())
    if (GV.getValueType() == OpenMPIRBuilder::Ident && GV.hasInitializer())
      if (GV.getInitializer() == Initializer)
        Ident = &GV;

  if (!Ident) {
    auto *GV = new GlobalVariable(
        M, OpenMPIRBuilder::Ident,
        /* isConstant = */ true, GlobalValue::PrivateLinkage, Initializer, "",
        nullptr, GlobalValue::NotThreadLocal,
        M.getDataLayout().getDefaultGlobalsAddressSpace());
    GV->setUnnamedAddr(GlobalValue::UnnamedAddr::Global);
    GV->setAlignment(Align(8));
    Ident = GV;
  }
}

return ConstantExpr::getPointerBitCastOrAddrSpaceCast(Ident, IdentPtr);
592}

594Constant *OpenMPIRBuilder::getOrCreateSrcLocStr(StringRef LocStr,
                                              uint32_t &SrcLocStrSize) {
SrcLocStrSize = LocStr.size();
Constant *&SrcLocStr = SrcLocStrMap[LocStr];
if (!SrcLocStr) {
  Constant *Initializer =
      ConstantDataArray::getString(M.getContext(), LocStr);

  // Look for existing encoding of the location, not needed but minimizes the
  // difference to the existing solution while we transition.
  for (GlobalVariable &GV : M.globals())
    if (GV.isConstant() && GV.hasInitializer() &&
        GV.getInitializer() == Initializer)
      return SrcLocStr = ConstantExpr::getPointerCast(&GV, Int8Ptr);

  SrcLocStr = Builder.CreateGlobalStringPtr(LocStr, /* Name */ "",
                                            /* AddressSpace */ 0, &M);
}
return SrcLocStr;
613}

615Constant *OpenMPIRBuilder::getOrCreateSrcLocStr(StringRef FunctionName,
                                              StringRef FileName,
                                              unsigned Line, unsigned Column,
                                              uint32_t &SrcLocStrSize) {
SmallString<128> Buffer;
Buffer.push_back(';');
Buffer.append(FileName);
Buffer.push_back(';');
Buffer.append(FunctionName);
Buffer.push_back(';');
Buffer.append(std::to_string(Line));
Buffer.push_back(';');
Buffer.append(std::to_string(Column));
Buffer.push_back(';');
Buffer.push_back(';');
return getOrCreateSrcLocStr(Buffer.str(), SrcLocStrSize);
631}

633Constant *
634OpenMPIRBuilder::getOrCreateDefaultSrcLocStr(uint32_t &SrcLocStrSize) {
StringRef UnknownLoc = ";unknown;unknown;0;0;;";
return getOrCreateSrcLocStr(UnknownLoc, SrcLocStrSize);
637}

639Constant *OpenMPIRBuilder::getOrCreateSrcLocStr(DebugLoc DL,
                                              uint32_t &SrcLocStrSize,
                                              Function *F) {
DILocation *DIL = DL.get();
if (!DIL)
  return getOrCreateDefaultSrcLocStr(SrcLocStrSize);
StringRef FileName = M.getName();
if (DIFile *DIF = DIL->getFile())
  if (std::optional<StringRef> Source = DIF->getSource())
    FileName = *Source;
StringRef Function = DIL->getScope()->getSubprogram()->getName();
if (Function.empty() && F)
  Function = F->getName();
return getOrCreateSrcLocStr(Function, FileName, DIL->getLine(),
                            DIL->getColumn(), SrcLocStrSize);
654}

656Constant *OpenMPIRBuilder::getOrCreateSrcLocStr(const LocationDescription &Loc,
                                              uint32_t &SrcLocStrSize) {
return getOrCreateSrcLocStr(Loc.DL, SrcLocStrSize,
                            Loc.IP.getBlock()->getParent());
660}

662Value *OpenMPIRBuilder::getOrCreateThreadID(Value *Ident) {
return Builder.CreateCall(
    getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_global_thread_num), Ident,
    "omp_global_thread_num");
666}

668OpenMPIRBuilder::InsertPointTy
669OpenMPIRBuilder::createBarrier(const LocationDescription &Loc, Directive DK,
                             bool ForceSimpleCall, bool CheckCancelFlag) {
if (!updateToLocation(Loc))
  return Loc.IP;
return emitBarrierImpl(Loc, DK, ForceSimpleCall, CheckCancelFlag);
674}

676OpenMPIRBuilder::InsertPointTy
677OpenMPIRBuilder::emitBarrierImpl(const LocationDescription &Loc, Directive Kind,
                               bool ForceSimpleCall, bool CheckCancelFlag) {
// Build call __kmpc_cancel_barrier(loc, thread_id) or
//            __kmpc_barrier(loc, thread_id);

IdentFlag BarrierLocFlags;
switch (Kind) {
case OMPD_for:
  BarrierLocFlags = OMP_IDENT_FLAG_BARRIER_IMPL_FOR;
  break;
case OMPD_sections:
  BarrierLocFlags = OMP_IDENT_FLAG_BARRIER_IMPL_SECTIONS;
  break;
case OMPD_single:
  BarrierLocFlags = OMP_IDENT_FLAG_BARRIER_IMPL_SINGLE;
  break;
case OMPD_barrier:
  BarrierLocFlags = OMP_IDENT_FLAG_BARRIER_EXPL;
  break;
default:
  BarrierLocFlags = OMP_IDENT_FLAG_BARRIER_IMPL;
  break;
}

uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Args[] = {
    getOrCreateIdent(SrcLocStr, SrcLocStrSize, BarrierLocFlags),
    getOrCreateThreadID(getOrCreateIdent(SrcLocStr, SrcLocStrSize))};

// If we are in a cancellable parallel region, barriers are cancellation
// points.
// TODO: Check why we would force simple calls or to ignore the cancel flag.
bool UseCancelBarrier =
    !ForceSimpleCall && isLastFinalizationInfoCancellable(OMPD_parallel);

Value *Result =
    Builder.CreateCall(getOrCreateRuntimeFunctionPtr(
                           UseCancelBarrier ? OMPRTL___kmpc_cancel_barrier
                                            : OMPRTL___kmpc_barrier),
                       Args);

if (UseCancelBarrier && CheckCancelFlag)
  emitCancelationCheckImpl(Result, OMPD_parallel);

return Builder.saveIP();
723}

725OpenMPIRBuilder::InsertPointTy
726OpenMPIRBuilder::createCancel(const LocationDescription &Loc,
                            Value *IfCondition,
                            omp::Directive CanceledDirective) {
if (!updateToLocation(Loc))
  return Loc.IP;

// LLVM utilities like blocks with terminators.
auto *UI = Builder.CreateUnreachable();

Instruction *ThenTI = UI, *ElseTI = nullptr;
if (IfCondition)
  SplitBlockAndInsertIfThenElse(IfCondition, UI, &ThenTI, &ElseTI);
Builder.SetInsertPoint(ThenTI);

Value *CancelKind = nullptr;
switch (CanceledDirective) {
742#define OMP_CANCEL_KIND(Enum, Str, DirectiveEnum, Value)                       \
case DirectiveEnum:                                                          \
  CancelKind = Builder.getInt32(Value);                                      \
  break;
746#include "llvm/Frontend/OpenMP/OMPKinds.def"
default:
  llvm_unreachable("Unknown cancel kind!")::llvm::llvm_unreachable_internal("Unknown cancel kind!", "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp"
, 748);
}

uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
Value *Args[] = {Ident, getOrCreateThreadID(Ident), CancelKind};
Value *Result = Builder.CreateCall(
    getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_cancel), Args);
auto ExitCB = [this, CanceledDirective, Loc](InsertPointTy IP) {
  if (CanceledDirective == OMPD_parallel) {
    IRBuilder<>::InsertPointGuard IPG(Builder);
    Builder.restoreIP(IP);
    createBarrier(LocationDescription(Builder.saveIP(), Loc.DL),
                  omp::Directive::OMPD_unknown, /* ForceSimpleCall */ false,
                  /* CheckCancelFlag */ false);
  }
};

// The actual cancel logic is shared with others, e.g., cancel_barriers.
emitCancelationCheckImpl(Result, CanceledDirective, ExitCB);

// Update the insertion point and remove the terminator we introduced.
Builder.SetInsertPoint(UI->getParent());
UI->eraseFromParent();

return Builder.saveIP();
775}

777void OpenMPIRBuilder::emitOffloadingEntry(Constant *Addr, StringRef Name,
                                        uint64_t Size, int32_t Flags,
                                        StringRef SectionName) {
Type *Int8PtrTy = Type::getInt8PtrTy(M.getContext());
Type *Int32Ty = Type::getInt32Ty(M.getContext());
Type *SizeTy = M.getDataLayout().getIntPtrType(M.getContext());

Constant *AddrName = ConstantDataArray::getString(M.getContext(), Name);

// Create the constant string used to look up the symbol in the device.
auto *Str =
    new llvm::GlobalVariable(M, AddrName->getType(), /*isConstant=*/true,
                             llvm::GlobalValue::InternalLinkage, AddrName,
                             ".omp_offloading.entry_name");
Str->setUnnamedAddr(llvm::GlobalValue::UnnamedAddr::Global);

// Construct the offloading entry.
Constant *EntryData[] = {
    ConstantExpr::getPointerBitCastOrAddrSpaceCast(Addr, Int8PtrTy),
    ConstantExpr::getPointerBitCastOrAddrSpaceCast(Str, Int8PtrTy),
    ConstantInt::get(SizeTy, Size),
    ConstantInt::get(Int32Ty, Flags),
    ConstantInt::get(Int32Ty, 0),
};
Constant *EntryInitializer =
    ConstantStruct::get(OpenMPIRBuilder::OffloadEntry, EntryData);

auto *Entry = new GlobalVariable(
    M, OpenMPIRBuilder::OffloadEntry,
    /* isConstant = */ true, GlobalValue::WeakAnyLinkage, EntryInitializer,
    ".omp_offloading.entry." + Name, nullptr, GlobalValue::NotThreadLocal,
    M.getDataLayout().getDefaultGlobalsAddressSpace());

// The entry has to be created in the section the linker expects it to be.
Entry->setSection(SectionName);
Entry->setAlignment(Align(1));
813}

815OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::emitTargetKernel(
  const LocationDescription &Loc, InsertPointTy AllocaIP, Value *&Return,
  Value *Ident, Value *DeviceID, Value *NumTeams, Value *NumThreads,
  Value *HostPtr, ArrayRef<Value *> KernelArgs) {
if (!updateToLocation(Loc))
  return Loc.IP;

Builder.restoreIP(AllocaIP);
auto *KernelArgsPtr =
    Builder.CreateAlloca(OpenMPIRBuilder::KernelArgs, nullptr, "kernel_args");
Builder.restoreIP(Loc.IP);

for (unsigned I = 0, Size = KernelArgs.size(); I != Size; ++I) {
  llvm::Value *Arg =
      Builder.CreateStructGEP(OpenMPIRBuilder::KernelArgs, KernelArgsPtr, I);
  Builder.CreateAlignedStore(
      KernelArgs[I], Arg,
      M.getDataLayout().getPrefTypeAlign(KernelArgs[I]->getType()));
}

SmallVector<Value *> OffloadingArgs{Ident,      DeviceID, NumTeams,
                                    NumThreads, HostPtr,  KernelArgsPtr};

Return = Builder.CreateCall(
    getOrCreateRuntimeFunction(M, OMPRTL___tgt_target_kernel),
    OffloadingArgs);

return Builder.saveIP();
843}

845void OpenMPIRBuilder::emitCancelationCheckImpl(Value *CancelFlag,
                                             omp::Directive CanceledDirective,
                                             FinalizeCallbackTy ExitCB) {
assert(isLastFinalizationInfoCancellable(CanceledDirective) &&(static_cast <bool> (isLastFinalizationInfoCancellable(
CanceledDirective) && "Unexpected cancellation!") ? void
 (0) : __assert_fail ("isLastFinalizationInfoCancellable(CanceledDirective) && \"Unexpected cancellation!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 849, __extension__
 __PRETTY_FUNCTION__))
       "Unexpected cancellation!")(static_cast <bool> (isLastFinalizationInfoCancellable(
CanceledDirective) && "Unexpected cancellation!") ? void
 (0) : __assert_fail ("isLastFinalizationInfoCancellable(CanceledDirective) && \"Unexpected cancellation!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 849, __extension__
 __PRETTY_FUNCTION__));

// For a cancel barrier we create two new blocks.
BasicBlock *BB = Builder.GetInsertBlock();
BasicBlock *NonCancellationBlock;
if (Builder.GetInsertPoint() == BB->end()) {
  // TODO: This branch will not be needed once we moved to the
  // OpenMPIRBuilder codegen completely.
  NonCancellationBlock = BasicBlock::Create(
      BB->getContext(), BB->getName() + ".cont", BB->getParent());
} else {
  NonCancellationBlock = SplitBlock(BB, &*Builder.GetInsertPoint());
  BB->getTerminator()->eraseFromParent();
  Builder.SetInsertPoint(BB);
}
BasicBlock *CancellationBlock = BasicBlock::Create(
    BB->getContext(), BB->getName() + ".cncl", BB->getParent());

// Jump to them based on the return value.
Value *Cmp = Builder.CreateIsNull(CancelFlag);
Builder.CreateCondBr(Cmp, NonCancellationBlock, CancellationBlock,
                     /* TODO weight */ nullptr, nullptr);

// From the cancellation block we finalize all variables and go to the
// post finalization block that is known to the FiniCB callback.
Builder.SetInsertPoint(CancellationBlock);
if (ExitCB)
  ExitCB(Builder.saveIP());
auto &FI = FinalizationStack.back();
FI.FiniCB(Builder.saveIP());

// The continuation block is where code generation continues.
Builder.SetInsertPoint(NonCancellationBlock, NonCancellationBlock->begin());
882}

884IRBuilder<>::InsertPoint OpenMPIRBuilder::createParallel(
  const LocationDescription &Loc, InsertPointTy OuterAllocaIP,
  BodyGenCallbackTy BodyGenCB, PrivatizeCallbackTy PrivCB,
  FinalizeCallbackTy FiniCB, Value *IfCondition, Value *NumThreads,
  omp::ProcBindKind ProcBind, bool IsCancellable) {
assert(!isConflictIP(Loc.IP, OuterAllocaIP) && "IPs must not be ambiguous")(static_cast <bool> (!isConflictIP(Loc.IP, OuterAllocaIP
) && "IPs must not be ambiguous") ? void (0) : __assert_fail
 ("!isConflictIP(Loc.IP, OuterAllocaIP) && \"IPs must not be ambiguous\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 889, __extension__
 __PRETTY_FUNCTION__));

if (!updateToLocation(Loc))
  return Loc.IP;

uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
Value *ThreadID = getOrCreateThreadID(Ident);

if (NumThreads) {
  // Build call __kmpc_push_num_threads(&Ident, global_tid, num_threads)
  Value *Args[] = {
      Ident, ThreadID,
      Builder.CreateIntCast(NumThreads, Int32, /*isSigned*/ false)};
  Builder.CreateCall(
      getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_push_num_threads), Args);
}

if (ProcBind != OMP_PROC_BIND_default) {
  // Build call __kmpc_push_proc_bind(&Ident, global_tid, proc_bind)
  Value *Args[] = {
      Ident, ThreadID,
      ConstantInt::get(Int32, unsigned(ProcBind), /*isSigned=*/true)};
  Builder.CreateCall(
      getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_push_proc_bind), Args);
}

BasicBlock *InsertBB = Builder.GetInsertBlock();
Function *OuterFn = InsertBB->getParent();

// Save the outer alloca block because the insertion iterator may get
// invalidated and we still need this later.
BasicBlock *OuterAllocaBlock = OuterAllocaIP.getBlock();

// Vector to remember instructions we used only during the modeling but which
// we want to delete at the end.
SmallVector<Instruction *, 4> ToBeDeleted;

// Change the location to the outer alloca insertion point to create and
// initialize the allocas we pass into the parallel region.
Builder.restoreIP(OuterAllocaIP);
AllocaInst *TIDAddr = Builder.CreateAlloca(Int32, nullptr, "tid.addr");
AllocaInst *ZeroAddr = Builder.CreateAlloca(Int32, nullptr, "zero.addr");

// We only need TIDAddr and ZeroAddr for modeling purposes to get the
// associated arguments in the outlined function, so we delete them later.
ToBeDeleted.push_back(TIDAddr);
ToBeDeleted.push_back(ZeroAddr);

// Create an artificial insertion point that will also ensure the blocks we
// are about to split are not degenerated.
auto *UI = new UnreachableInst(Builder.getContext(), InsertBB);

BasicBlock *EntryBB = UI->getParent();
BasicBlock *PRegEntryBB = EntryBB->splitBasicBlock(UI, "omp.par.entry");
BasicBlock *PRegBodyBB = PRegEntryBB->splitBasicBlock(UI, "omp.par.region");
BasicBlock *PRegPreFiniBB =
    PRegBodyBB->splitBasicBlock(UI, "omp.par.pre_finalize");
BasicBlock *PRegExitBB = PRegPreFiniBB->splitBasicBlock(UI, "omp.par.exit");

auto FiniCBWrapper = [&](InsertPointTy IP) {
  // Hide "open-ended" blocks from the given FiniCB by setting the right jump
  // target to the region exit block.
  if (IP.getBlock()->end() == IP.getPoint()) {
    IRBuilder<>::InsertPointGuard IPG(Builder);
    Builder.restoreIP(IP);
    Instruction *I = Builder.CreateBr(PRegExitBB);
    IP = InsertPointTy(I->getParent(), I->getIterator());
  }
  assert(IP.getBlock()->getTerminator()->getNumSuccessors() == 1 &&(static_cast <bool> (IP.getBlock()->getTerminator()->
getNumSuccessors() == 1 && IP.getBlock()->getTerminator
()->getSuccessor(0) == PRegExitBB && "Unexpected insertion point for finalization call!"
) ? void (0) : __assert_fail ("IP.getBlock()->getTerminator()->getNumSuccessors() == 1 && IP.getBlock()->getTerminator()->getSuccessor(0) == PRegExitBB && \"Unexpected insertion point for finalization call!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 961, __extension__
 __PRETTY_FUNCTION__))
         IP.getBlock()->getTerminator()->getSuccessor(0) == PRegExitBB &&(static_cast <bool> (IP.getBlock()->getTerminator()->
getNumSuccessors() == 1 && IP.getBlock()->getTerminator
()->getSuccessor(0) == PRegExitBB && "Unexpected insertion point for finalization call!"
) ? void (0) : __assert_fail ("IP.getBlock()->getTerminator()->getNumSuccessors() == 1 && IP.getBlock()->getTerminator()->getSuccessor(0) == PRegExitBB && \"Unexpected insertion point for finalization call!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 961, __extension__
 __PRETTY_FUNCTION__))
         "Unexpected insertion point for finalization call!")(static_cast <bool> (IP.getBlock()->getTerminator()->
getNumSuccessors() == 1 && IP.getBlock()->getTerminator
()->getSuccessor(0) == PRegExitBB && "Unexpected insertion point for finalization call!"
) ? void (0) : __assert_fail ("IP.getBlock()->getTerminator()->getNumSuccessors() == 1 && IP.getBlock()->getTerminator()->getSuccessor(0) == PRegExitBB && \"Unexpected insertion point for finalization call!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 961, __extension__
 __PRETTY_FUNCTION__));
  return FiniCB(IP);
};

FinalizationStack.push_back({FiniCBWrapper, OMPD_parallel, IsCancellable});

// Generate the privatization allocas in the block that will become the entry
// of the outlined function.
Builder.SetInsertPoint(PRegEntryBB->getTerminator());
InsertPointTy InnerAllocaIP = Builder.saveIP();

AllocaInst *PrivTIDAddr =
    Builder.CreateAlloca(Int32, nullptr, "tid.addr.local");
Instruction *PrivTID = Builder.CreateLoad(Int32, PrivTIDAddr, "tid");

// Add some fake uses for OpenMP provided arguments.
ToBeDeleted.push_back(Builder.CreateLoad(Int32, TIDAddr, "tid.addr.use"));
Instruction *ZeroAddrUse =
    Builder.CreateLoad(Int32, ZeroAddr, "zero.addr.use");
ToBeDeleted.push_back(ZeroAddrUse);

// EntryBB
//   |
//   V
// PRegionEntryBB         <- Privatization allocas are placed here.
//   |
//   V
// PRegionBodyBB          <- BodeGen is invoked here.
//   |
//   V
// PRegPreFiniBB          <- The block we will start finalization from.
//   |
//   V
// PRegionExitBB          <- A common exit to simplify block collection.
//

LLVM_DEBUG(dbgs() << "Before body codegen: " << *OuterFn << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Before body codegen: "
 << *OuterFn << "\n"; } } while (false);

// Let the caller create the body.
assert(BodyGenCB && "Expected body generation callback!")(static_cast <bool> (BodyGenCB && "Expected body generation callback!"
) ? void (0) : __assert_fail ("BodyGenCB && \"Expected body generation callback!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1000, __extension__
 __PRETTY_FUNCTION__));
InsertPointTy CodeGenIP(PRegBodyBB, PRegBodyBB->begin());
BodyGenCB(InnerAllocaIP, CodeGenIP);

LLVM_DEBUG(dbgs() << "After  body codegen: " << *OuterFn << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "After  body codegen: "
 << *OuterFn << "\n"; } } while (false);
FunctionCallee RTLFn;
if (IfCondition)
  RTLFn = getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_fork_call_if);
else
  RTLFn = getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_fork_call);

if (auto *F = dyn_cast<llvm::Function>(RTLFn.getCallee())) {
  if (!F->hasMetadata(llvm::LLVMContext::MD_callback)) {
    llvm::LLVMContext &Ctx = F->getContext();
    MDBuilder MDB(Ctx);
    // Annotate the callback behavior of the __kmpc_fork_call:
    //  - The callback callee is argument number 2 (microtask).
    //  - The first two arguments of the callback callee are unknown (-1).
    //  - All variadic arguments to the __kmpc_fork_call are passed to the
    //    callback callee.
    F->addMetadata(
        llvm::LLVMContext::MD_callback,
        *llvm::MDNode::get(
            Ctx, {MDB.createCallbackEncoding(2, {-1, -1},
                                             /* VarArgsArePassed */ true)}));
  }
}

OutlineInfo OI;
OI.PostOutlineCB = [=](Function &OutlinedFn) {
  // Add some known attributes.
  OutlinedFn.addParamAttr(0, Attribute::NoAlias);
  OutlinedFn.addParamAttr(1, Attribute::NoAlias);
  OutlinedFn.addFnAttr(Attribute::NoUnwind);
  OutlinedFn.addFnAttr(Attribute::NoRecurse);

  assert(OutlinedFn.arg_size() >= 2 &&(static_cast <bool> (OutlinedFn.arg_size() >= 2 &&
 "Expected at least tid and bounded tid as arguments") ? void
 (0) : __assert_fail ("OutlinedFn.arg_size() >= 2 && \"Expected at least tid and bounded tid as arguments\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1037, __extension__
 __PRETTY_FUNCTION__))
         "Expected at least tid and bounded tid as arguments")(static_cast <bool> (OutlinedFn.arg_size() >= 2 &&
 "Expected at least tid and bounded tid as arguments") ? void
 (0) : __assert_fail ("OutlinedFn.arg_size() >= 2 && \"Expected at least tid and bounded tid as arguments\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1037, __extension__
 __PRETTY_FUNCTION__));
  unsigned NumCapturedVars =
      OutlinedFn.arg_size() - /* tid & bounded tid */ 2;

  CallInst *CI = cast<CallInst>(OutlinedFn.user_back());
  CI->getParent()->setName("omp_parallel");
  Builder.SetInsertPoint(CI);

  // Build call __kmpc_fork_call[_if](Ident, n, microtask, var1, .., varn);
  Value *ForkCallArgs[] = {
      Ident, Builder.getInt32(NumCapturedVars),
      Builder.CreateBitCast(&OutlinedFn, ParallelTaskPtr)};

  SmallVector<Value *, 16> RealArgs;
  RealArgs.append(std::begin(ForkCallArgs), std::end(ForkCallArgs));
  if (IfCondition) {
    Value *Cond = Builder.CreateSExtOrTrunc(IfCondition,
                                            Type::getInt32Ty(M.getContext()));
    RealArgs.push_back(Cond);
  }
  RealArgs.append(CI->arg_begin() + /* tid & bound tid */ 2, CI->arg_end());

  // __kmpc_fork_call_if always expects a void ptr as the last argument
  // If there are no arguments, pass a null pointer.
  auto PtrTy = Type::getInt8PtrTy(M.getContext());
  if (IfCondition && NumCapturedVars == 0) {
    llvm::Value *Void = ConstantPointerNull::get(PtrTy);
    RealArgs.push_back(Void);
  }
  if (IfCondition && RealArgs.back()->getType() != PtrTy)
    RealArgs.back() = Builder.CreateBitCast(RealArgs.back(), PtrTy);

  Builder.CreateCall(RTLFn, RealArgs);

  LLVM_DEBUG(dbgs() << "With fork_call placed: "do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "With fork_call placed: "
 << *Builder.GetInsertBlock()->getParent() << "\n"
; } } while (false)
                    << *Builder.GetInsertBlock()->getParent() << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "With fork_call placed: "
 << *Builder.GetInsertBlock()->getParent() << "\n"
; } } while (false);

  InsertPointTy ExitIP(PRegExitBB, PRegExitBB->end());

  // Initialize the local TID stack location with the argument value.
  Builder.SetInsertPoint(PrivTID);
  Function::arg_iterator OutlinedAI = OutlinedFn.arg_begin();
  Builder.CreateStore(Builder.CreateLoad(Int32, OutlinedAI), PrivTIDAddr);

  CI->eraseFromParent();

  for (Instruction *I : ToBeDeleted)
    I->eraseFromParent();
};

// Adjust the finalization stack, verify the adjustment, and call the
// finalize function a last time to finalize values between the pre-fini
// block and the exit block if we left the parallel "the normal way".
auto FiniInfo = FinalizationStack.pop_back_val();
(void)FiniInfo;
assert(FiniInfo.DK == OMPD_parallel &&(static_cast <bool> (FiniInfo.DK == OMPD_parallel &&
 "Unexpected finalization stack state!") ? void (0) : __assert_fail
 ("FiniInfo.DK == OMPD_parallel && \"Unexpected finalization stack state!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1093, __extension__
 __PRETTY_FUNCTION__))
       "Unexpected finalization stack state!")(static_cast <bool> (FiniInfo.DK == OMPD_parallel &&
 "Unexpected finalization stack state!") ? void (0) : __assert_fail
 ("FiniInfo.DK == OMPD_parallel && \"Unexpected finalization stack state!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1093, __extension__
 __PRETTY_FUNCTION__));

Instruction *PRegPreFiniTI = PRegPreFiniBB->getTerminator();

InsertPointTy PreFiniIP(PRegPreFiniBB, PRegPreFiniTI->getIterator());
FiniCB(PreFiniIP);

OI.OuterAllocaBB = OuterAllocaBlock;
OI.EntryBB = PRegEntryBB;
OI.ExitBB = PRegExitBB;

SmallPtrSet<BasicBlock *, 32> ParallelRegionBlockSet;
SmallVector<BasicBlock *, 32> Blocks;
OI.collectBlocks(ParallelRegionBlockSet, Blocks);

// Ensure a single exit node for the outlined region by creating one.
// We might have multiple incoming edges to the exit now due to finalizations,
// e.g., cancel calls that cause the control flow to leave the region.
BasicBlock *PRegOutlinedExitBB = PRegExitBB;
PRegExitBB = SplitBlock(PRegExitBB, &*PRegExitBB->getFirstInsertionPt());
PRegOutlinedExitBB->setName("omp.par.outlined.exit");
Blocks.push_back(PRegOutlinedExitBB);

CodeExtractorAnalysisCache CEAC(*OuterFn);
CodeExtractor Extractor(Blocks, /* DominatorTree */ nullptr,
                        /* AggregateArgs */ false,
                        /* BlockFrequencyInfo */ nullptr,
                        /* BranchProbabilityInfo */ nullptr,
                        /* AssumptionCache */ nullptr,
                        /* AllowVarArgs */ true,
                        /* AllowAlloca */ true,
                        /* AllocationBlock */ OuterAllocaBlock,
                        /* Suffix */ ".omp_par");

// Find inputs to, outputs from the code region.
BasicBlock *CommonExit = nullptr;
SetVector<Value *> Inputs, Outputs, SinkingCands, HoistingCands;
Extractor.findAllocas(CEAC, SinkingCands, HoistingCands, CommonExit);
Extractor.findInputsOutputs(Inputs, Outputs, SinkingCands);

LLVM_DEBUG(dbgs() << "Before privatization: " << *OuterFn << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Before privatization: "
 << *OuterFn << "\n"; } } while (false);

FunctionCallee TIDRTLFn =
    getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_global_thread_num);

auto PrivHelper = [&](Value &V) {
  if (&V == TIDAddr || &V == ZeroAddr) {
    OI.ExcludeArgsFromAggregate.push_back(&V);
    return;
  }

  SetVector<Use *> Uses;
  for (Use &U : V.uses())
    if (auto *UserI = dyn_cast<Instruction>(U.getUser()))
      if (ParallelRegionBlockSet.count(UserI->getParent()))
        Uses.insert(&U);

  // __kmpc_fork_call expects extra arguments as pointers. If the input
  // already has a pointer type, everything is fine. Otherwise, store the
  // value onto stack and load it back inside the to-be-outlined region. This
  // will ensure only the pointer will be passed to the function.
  // FIXME: if there are more than 15 trailing arguments, they must be
  // additionally packed in a struct.
  Value *Inner = &V;
  if (!V.getType()->isPointerTy()) {
    IRBuilder<>::InsertPointGuard Guard(Builder);
    LLVM_DEBUG(llvm::dbgs() << "Forwarding input as pointer: " << V << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { llvm::dbgs() << "Forwarding input as pointer: "
 << V << "\n"; } } while (false);

    Builder.restoreIP(OuterAllocaIP);
    Value *Ptr =
        Builder.CreateAlloca(V.getType(), nullptr, V.getName() + ".reloaded");

    // Store to stack at end of the block that currently branches to the entry
    // block of the to-be-outlined region.
    Builder.SetInsertPoint(InsertBB,
                           InsertBB->getTerminator()->getIterator());
    Builder.CreateStore(&V, Ptr);

    // Load back next to allocations in the to-be-outlined region.
    Builder.restoreIP(InnerAllocaIP);
    Inner = Builder.CreateLoad(V.getType(), Ptr);
  }

  Value *ReplacementValue = nullptr;
  CallInst *CI = dyn_cast<CallInst>(&V);
  if (CI && CI->getCalledFunction() == TIDRTLFn.getCallee()) {
    ReplacementValue = PrivTID;
  } else {
    Builder.restoreIP(
        PrivCB(InnerAllocaIP, Builder.saveIP(), V, *Inner, ReplacementValue));
    assert(ReplacementValue &&(static_cast <bool> (ReplacementValue && "Expected copy/create callback to set replacement value!"
) ? void (0) : __assert_fail ("ReplacementValue && \"Expected copy/create callback to set replacement value!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1184, __extension__
 __PRETTY_FUNCTION__))
           "Expected copy/create callback to set replacement value!")(static_cast <bool> (ReplacementValue && "Expected copy/create callback to set replacement value!"
) ? void (0) : __assert_fail ("ReplacementValue && \"Expected copy/create callback to set replacement value!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1184, __extension__
 __PRETTY_FUNCTION__));
    if (ReplacementValue == &V)
      return;
  }

  for (Use *UPtr : Uses)
    UPtr->set(ReplacementValue);
};

// Reset the inner alloca insertion as it will be used for loading the values
// wrapped into pointers before passing them into the to-be-outlined region.
// Configure it to insert immediately after the fake use of zero address so
// that they are available in the generated body and so that the
// OpenMP-related values (thread ID and zero address pointers) remain leading
// in the argument list.
InnerAllocaIP = IRBuilder<>::InsertPoint(
    ZeroAddrUse->getParent(), ZeroAddrUse->getNextNode()->getIterator());

// Reset the outer alloca insertion point to the entry of the relevant block
// in case it was invalidated.
OuterAllocaIP = IRBuilder<>::InsertPoint(
    OuterAllocaBlock, OuterAllocaBlock->getFirstInsertionPt());

for (Value *Input : Inputs) {
  LLVM_DEBUG(dbgs() << "Captured input: " << *Input << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Captured input: " <<
 *Input << "\n"; } } while (false);
  PrivHelper(*Input);
}
LLVM_DEBUG({do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { for (Value *Output : Outputs) do {
 if (::llvm::DebugFlag && ::llvm::isCurrentDebugType(
"openmp-ir-builder")) { dbgs() << "Captured output: " <<
 *Output << "\n"; } } while (false); }; } } while (false
)
  for (Value *Output : Outputs)do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { for (Value *Output : Outputs) do {
 if (::llvm::DebugFlag && ::llvm::isCurrentDebugType(
"openmp-ir-builder")) { dbgs() << "Captured output: " <<
 *Output << "\n"; } } while (false); }; } } while (false
)
    LLVM_DEBUG(dbgs() << "Captured output: " << *Output << "\n");do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { for (Value *Output : Outputs) do {
 if (::llvm::DebugFlag && ::llvm::isCurrentDebugType(
"openmp-ir-builder")) { dbgs() << "Captured output: " <<
 *Output << "\n"; } } while (false); }; } } while (false
)
})do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { for (Value *Output : Outputs) do {
 if (::llvm::DebugFlag && ::llvm::isCurrentDebugType(
"openmp-ir-builder")) { dbgs() << "Captured output: " <<
 *Output << "\n"; } } while (false); }; } } while (false
);
assert(Outputs.empty() &&(static_cast <bool> (Outputs.empty() && "OpenMP outlining should not produce live-out values!"
) ? void (0) : __assert_fail ("Outputs.empty() && \"OpenMP outlining should not produce live-out values!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1216, __extension__
 __PRETTY_FUNCTION__))
       "OpenMP outlining should not produce live-out values!")(static_cast <bool> (Outputs.empty() && "OpenMP outlining should not produce live-out values!"
) ? void (0) : __assert_fail ("Outputs.empty() && \"OpenMP outlining should not produce live-out values!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1216, __extension__
 __PRETTY_FUNCTION__));

LLVM_DEBUG(dbgs() << "After  privatization: " << *OuterFn << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "After  privatization: "
 << *OuterFn << "\n"; } } while (false);
LLVM_DEBUG({do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { for (auto *BB : Blocks) dbgs() <<
 " PBR: " << BB->getName() << "\n"; }; } } while
 (false)
  for (auto *BB : Blocks)do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { for (auto *BB : Blocks) dbgs() <<
 " PBR: " << BB->getName() << "\n"; }; } } while
 (false)
    dbgs() << " PBR: " << BB->getName() << "\n";do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { for (auto *BB : Blocks) dbgs() <<
 " PBR: " << BB->getName() << "\n"; }; } } while
 (false)
})do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { for (auto *BB : Blocks) dbgs() <<
 " PBR: " << BB->getName() << "\n"; }; } } while
 (false);

// Register the outlined info.
addOutlineInfo(std::move(OI));

InsertPointTy AfterIP(UI->getParent(), UI->getParent()->end());
UI->eraseFromParent();

return AfterIP;
1231}

1233void OpenMPIRBuilder::emitFlush(const LocationDescription &Loc) {
// Build call void __kmpc_flush(ident_t *loc)
uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Args[] = {getOrCreateIdent(SrcLocStr, SrcLocStrSize)};

Builder.CreateCall(getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_flush), Args);
1240}

1242void OpenMPIRBuilder::createFlush(const LocationDescription &Loc) {
if (!updateToLocation(Loc))
  return;
emitFlush(Loc);
1246}

1248void OpenMPIRBuilder::emitTaskwaitImpl(const LocationDescription &Loc) {
// Build call kmp_int32 __kmpc_omp_taskwait(ident_t *loc, kmp_int32
// global_tid);
uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
Value *Args[] = {Ident, getOrCreateThreadID(Ident)};

// Ignore return result until untied tasks are supported.
Builder.CreateCall(getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_omp_taskwait),
                   Args);
1259}

1261void OpenMPIRBuilder::createTaskwait(const LocationDescription &Loc) {
if (!updateToLocation(Loc))
  return;
emitTaskwaitImpl(Loc);
1265}

1267void OpenMPIRBuilder::emitTaskyieldImpl(const LocationDescription &Loc) {
// Build call __kmpc_omp_taskyield(loc, thread_id, 0);
uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
Constant *I32Null = ConstantInt::getNullValue(Int32);
Value *Args[] = {Ident, getOrCreateThreadID(Ident), I32Null};

Builder.CreateCall(getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_omp_taskyield),
                   Args);
1277}

1279void OpenMPIRBuilder::createTaskyield(const LocationDescription &Loc) {
if (!updateToLocation(Loc))
  return;
emitTaskyieldImpl(Loc);
1283}

1285OpenMPIRBuilder::InsertPointTy
1286OpenMPIRBuilder::createTask(const LocationDescription &Loc,
                          InsertPointTy AllocaIP, BodyGenCallbackTy BodyGenCB,
                          bool Tied, Value *Final, Value *IfCondition,
                          SmallVector<DependData> Dependencies) {
if (!updateToLocation(Loc))
  return InsertPointTy();

uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
// The current basic block is split into four basic blocks. After outlining,
// they will be mapped as follows:
// ```
// def current_fn() {
//   current_basic_block:
//     br label %task.exit
//   task.exit:
//     ; instructions after task
// }
// def outlined_fn() {
//   task.alloca:
//     br label %task.body
//   task.body:
//     ret void
// }
// ```
BasicBlock *TaskExitBB = splitBB(Builder, /*CreateBranch=*/true, "task.exit");
BasicBlock *TaskBodyBB = splitBB(Builder, /*CreateBranch=*/true, "task.body");
BasicBlock *TaskAllocaBB =
    splitBB(Builder, /*CreateBranch=*/true, "task.alloca");

OutlineInfo OI;
OI.EntryBB = TaskAllocaBB;
OI.OuterAllocaBB = AllocaIP.getBlock();
OI.ExitBB = TaskExitBB;
OI.PostOutlineCB = [this, Ident, Tied, Final, IfCondition,
                    Dependencies](Function &OutlinedFn) {
  // The input IR here looks like the following-
  // ```
  // func @current_fn() {
  //   outlined_fn(%args)
  // }
  // func @outlined_fn(%args) { ... }
  // ```
  //
  // This is changed to the following-
  //
  // ```
  // func @current_fn() {
  //   runtime_call(..., wrapper_fn, ...)
  // }
  // func @wrapper_fn(..., %args) {
  //   outlined_fn(%args)
  // }
  // func @outlined_fn(%args) { ... }
  // ```

  // The stale call instruction will be replaced with a new call instruction
  // for runtime call with a wrapper function.
  assert(OutlinedFn.getNumUses() == 1 &&(static_cast <bool> (OutlinedFn.getNumUses() == 1 &&
 "there must be a single user for the outlined function") ? void
 (0) : __assert_fail ("OutlinedFn.getNumUses() == 1 && \"there must be a single user for the outlined function\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1346, __extension__
 __PRETTY_FUNCTION__))
         "there must be a single user for the outlined function")(static_cast <bool> (OutlinedFn.getNumUses() == 1 &&
 "there must be a single user for the outlined function") ? void
 (0) : __assert_fail ("OutlinedFn.getNumUses() == 1 && \"there must be a single user for the outlined function\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1346, __extension__
 __PRETTY_FUNCTION__));
  CallInst *StaleCI = cast<CallInst>(OutlinedFn.user_back());

  // HasTaskData is true if any variables are captured in the outlined region,
  // false otherwise.
  bool HasTaskData = StaleCI->arg_size() > 0;
  Builder.SetInsertPoint(StaleCI);

  // Gather the arguments for emitting the runtime call for
  // @__kmpc_omp_task_alloc
  Function *TaskAllocFn =
      getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_omp_task_alloc);

  // Arguments - `loc_ref` (Ident) and `gtid` (ThreadID)
  // call.
  Value *ThreadID = getOrCreateThreadID(Ident);

  // Argument - `flags`
  // Task is tied iff (Flags & 1) == 1.
  // Task is untied iff (Flags & 1) == 0.
  // Task is final iff (Flags & 2) == 2.
  // Task is not final iff (Flags & 2) == 0.
  // TODO: Handle the other flags.
  Value *Flags = Builder.getInt32(Tied);
  if (Final) {
    Value *FinalFlag =
        Builder.CreateSelect(Final, Builder.getInt32(2), Builder.getInt32(0));
    Flags = Builder.CreateOr(FinalFlag, Flags);
  }

  // Argument - `sizeof_kmp_task_t` (TaskSize)
  // Tasksize refers to the size in bytes of kmp_task_t data structure
  // including private vars accessed in task.
  Value *TaskSize = Builder.getInt64(0);
  if (HasTaskData) {
    AllocaInst *ArgStructAlloca =
        dyn_cast<AllocaInst>(StaleCI->getArgOperand(0));
    assert(ArgStructAlloca &&(static_cast <bool> (ArgStructAlloca && "Unable to find the alloca instruction corresponding to arguments "
 "for extracted function") ? void (0) : __assert_fail ("ArgStructAlloca && \"Unable to find the alloca instruction corresponding to arguments \" \"for extracted function\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1385, __extension__
 __PRETTY_FUNCTION__))
           "Unable to find the alloca instruction corresponding to arguments "(static_cast <bool> (ArgStructAlloca && "Unable to find the alloca instruction corresponding to arguments "
 "for extracted function") ? void (0) : __assert_fail ("ArgStructAlloca && \"Unable to find the alloca instruction corresponding to arguments \" \"for extracted function\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1385, __extension__
 __PRETTY_FUNCTION__))
           "for extracted function")(static_cast <bool> (ArgStructAlloca && "Unable to find the alloca instruction corresponding to arguments "
 "for extracted function") ? void (0) : __assert_fail ("ArgStructAlloca && \"Unable to find the alloca instruction corresponding to arguments \" \"for extracted function\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1385, __extension__
 __PRETTY_FUNCTION__));
    StructType *ArgStructType =
        dyn_cast<StructType>(ArgStructAlloca->getAllocatedType());
    assert(ArgStructType && "Unable to find struct type corresponding to "(static_cast <bool> (ArgStructType && "Unable to find struct type corresponding to "
 "arguments for extracted function") ? void (0) : __assert_fail
 ("ArgStructType && \"Unable to find struct type corresponding to \" \"arguments for extracted function\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1389, __extension__
 __PRETTY_FUNCTION__))
                            "arguments for extracted function")(static_cast <bool> (ArgStructType && "Unable to find struct type corresponding to "
 "arguments for extracted function") ? void (0) : __assert_fail
 ("ArgStructType && \"Unable to find struct type corresponding to \" \"arguments for extracted function\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1389, __extension__
 __PRETTY_FUNCTION__));
    TaskSize =
        Builder.getInt64(M.getDataLayout().getTypeStoreSize(ArgStructType));
  }

  // TODO: Argument - sizeof_shareds

  // Argument - task_entry (the wrapper function)
  // If the outlined function has some captured variables (i.e. HasTaskData is
  // true), then the wrapper function will have an additional argument (the
  // struct containing captured variables). Otherwise, no such argument will
  // be present.
  SmallVector<Type *> WrapperArgTys{Builder.getInt32Ty()};
  if (HasTaskData)
    WrapperArgTys.push_back(OutlinedFn.getArg(0)->getType());
  FunctionCallee WrapperFuncVal = M.getOrInsertFunction(
      (Twine(OutlinedFn.getName()) + ".wrapper").str(),
      FunctionType::get(Builder.getInt32Ty(), WrapperArgTys, false));
  Function *WrapperFunc = dyn_cast<Function>(WrapperFuncVal.getCallee());
  PointerType *WrapperFuncBitcastType =
      FunctionType::get(Builder.getInt32Ty(),
                        {Builder.getInt32Ty(), Builder.getInt8PtrTy()}, false)
          ->getPointerTo();
  Value *WrapperFuncBitcast =
      ConstantExpr::getBitCast(WrapperFunc, WrapperFuncBitcastType);

  // Emit the @__kmpc_omp_task_alloc runtime call
  // The runtime call returns a pointer to an area where the task captured
  // variables must be copied before the task is run (NewTaskData)
  CallInst *NewTaskData = Builder.CreateCall(
      TaskAllocFn,
      {/*loc_ref=*/Ident, /*gtid=*/ThreadID, /*flags=*/Flags,
       /*sizeof_task=*/TaskSize, /*sizeof_shared=*/Builder.getInt64(0),
       /*task_func=*/WrapperFuncBitcast});

  // Copy the arguments for outlined function
  if (HasTaskData) {
    Value *TaskData = StaleCI->getArgOperand(0);
    Align Alignment = TaskData->getPointerAlignment(M.getDataLayout());
    Builder.CreateMemCpy(NewTaskData, Alignment, TaskData, Alignment,
                         TaskSize);
  }

  Value *DepArrayPtr = nullptr;
  if (Dependencies.size()) {
    InsertPointTy OldIP = Builder.saveIP();
    Builder.SetInsertPoint(
        &OldIP.getBlock()->getParent()->getEntryBlock().back());

    Type *DepArrayTy = ArrayType::get(DependInfo, Dependencies.size());
    Value *DepArray =
        Builder.CreateAlloca(DepArrayTy, nullptr, ".dep.arr.addr");

    unsigned P = 0;
    for (const DependData &Dep : Dependencies) {
      Value *Base =
          Builder.CreateConstInBoundsGEP2_64(DepArrayTy, DepArray, 0, P);
      // Store the pointer to the variable
      Value *Addr = Builder.CreateStructGEP(
          DependInfo, Base,
          static_cast<unsigned int>(RTLDependInfoFields::BaseAddr));
      Value *DepValPtr =
          Builder.CreatePtrToInt(Dep.DepVal, Builder.getInt64Ty());
      Builder.CreateStore(DepValPtr, Addr);
      // Store the size of the variable
      Value *Size = Builder.CreateStructGEP(
          DependInfo, Base,
          static_cast<unsigned int>(RTLDependInfoFields::Len));
      Builder.CreateStore(Builder.getInt64(M.getDataLayout().getTypeStoreSize(
                              Dep.DepValueType)),
                          Size);
      // Store the dependency kind
      Value *Flags = Builder.CreateStructGEP(
          DependInfo, Base,
          static_cast<unsigned int>(RTLDependInfoFields::Flags));
      Builder.CreateStore(
          ConstantInt::get(Builder.getInt8Ty(),
                           static_cast<unsigned int>(Dep.DepKind)),
          Flags);
      ++P;
    }

    DepArrayPtr = Builder.CreateBitCast(DepArray, Builder.getInt8PtrTy());
    Builder.restoreIP(OldIP);
  }

  // In the presence of the `if` clause, the following IR is generated:
  //    ...
  //    %data = call @__kmpc_omp_task_alloc(...)
  //    br i1 %if_condition, label %then, label %else
  //  then:
  //    call @__kmpc_omp_task(...)
  //    br label %exit
  //  else:
  //    call @__kmpc_omp_task_begin_if0(...)
  //    call @wrapper_fn(...)
  //    call @__kmpc_omp_task_complete_if0(...)
  //    br label %exit
  //  exit:
  //    ...
  if (IfCondition) {
    // `SplitBlockAndInsertIfThenElse` requires the block to have a
    // terminator.
    BasicBlock *NewBasicBlock =
        splitBB(Builder, /*CreateBranch=*/true, "if.end");
    Instruction *IfTerminator =
        NewBasicBlock->getSinglePredecessor()->getTerminator();
    Instruction *ThenTI = IfTerminator, *ElseTI = nullptr;
    Builder.SetInsertPoint(IfTerminator);
    SplitBlockAndInsertIfThenElse(IfCondition, IfTerminator, &ThenTI,
                                  &ElseTI);
    Builder.SetInsertPoint(ElseTI);
    Function *TaskBeginFn =
        getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_omp_task_begin_if0);
    Function *TaskCompleteFn =
        getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_omp_task_complete_if0);
    Builder.CreateCall(TaskBeginFn, {Ident, ThreadID, NewTaskData});
    if (HasTaskData)
      Builder.CreateCall(WrapperFunc, {ThreadID, NewTaskData});
    else
      Builder.CreateCall(WrapperFunc, {ThreadID});
    Builder.CreateCall(TaskCompleteFn, {Ident, ThreadID, NewTaskData});
    Builder.SetInsertPoint(ThenTI);
  }

  if (Dependencies.size()) {
    Function *TaskFn =
        getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_omp_task_with_deps);
    Builder.CreateCall(
        TaskFn,
        {Ident, ThreadID, NewTaskData, Builder.getInt32(Dependencies.size()),
         DepArrayPtr, ConstantInt::get(Builder.getInt32Ty(), 0),
         ConstantPointerNull::get(Type::getInt8PtrTy(M.getContext()))});

  } else {
    // Emit the @__kmpc_omp_task runtime call to spawn the task
    Function *TaskFn = getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_omp_task);
    Builder.CreateCall(TaskFn, {Ident, ThreadID, NewTaskData});
  }

  StaleCI->eraseFromParent();

  // Emit the body for wrapper function
  BasicBlock *WrapperEntryBB =
      BasicBlock::Create(M.getContext(), "", WrapperFunc);
  Builder.SetInsertPoint(WrapperEntryBB);
  if (HasTaskData)
    Builder.CreateCall(&OutlinedFn, {WrapperFunc->getArg(1)});
  else
    Builder.CreateCall(&OutlinedFn);
  Builder.CreateRet(Builder.getInt32(0));
};

addOutlineInfo(std::move(OI));

InsertPointTy TaskAllocaIP =
    InsertPointTy(TaskAllocaBB, TaskAllocaBB->begin());
InsertPointTy TaskBodyIP = InsertPointTy(TaskBodyBB, TaskBodyBB->begin());
BodyGenCB(TaskAllocaIP, TaskBodyIP);
Builder.SetInsertPoint(TaskExitBB, TaskExitBB->begin());

return Builder.saveIP();
1551}

1553OpenMPIRBuilder::InsertPointTy
1554OpenMPIRBuilder::createTaskgroup(const LocationDescription &Loc,
                               InsertPointTy AllocaIP,
                               BodyGenCallbackTy BodyGenCB) {
if (!updateToLocation(Loc))
  return InsertPointTy();

uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
Value *ThreadID = getOrCreateThreadID(Ident);

// Emit the @__kmpc_taskgroup runtime call to start the taskgroup
Function *TaskgroupFn =
    getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_taskgroup);
Builder.CreateCall(TaskgroupFn, {Ident, ThreadID});

BasicBlock *TaskgroupExitBB = splitBB(Builder, true, "taskgroup.exit");
BodyGenCB(AllocaIP, Builder.saveIP());

Builder.SetInsertPoint(TaskgroupExitBB);
// Emit the @__kmpc_end_taskgroup runtime call to end the taskgroup
Function *EndTaskgroupFn =
    getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_end_taskgroup);
Builder.CreateCall(EndTaskgroupFn, {Ident, ThreadID});

return Builder.saveIP();
1580}

1582OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::createSections(
  const LocationDescription &Loc, InsertPointTy AllocaIP,
  ArrayRef<StorableBodyGenCallbackTy> SectionCBs, PrivatizeCallbackTy PrivCB,
  FinalizeCallbackTy FiniCB, bool IsCancellable, bool IsNowait) {
assert(!isConflictIP(AllocaIP, Loc.IP) && "Dedicated IP allocas required")(static_cast <bool> (!isConflictIP(AllocaIP, Loc.IP) &&
 "Dedicated IP allocas required") ? void (0) : __assert_fail (
"!isConflictIP(AllocaIP, Loc.IP) && \"Dedicated IP allocas required\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1586, __extension__
 __PRETTY_FUNCTION__));

if (!updateToLocation(Loc))
  return Loc.IP;

auto FiniCBWrapper = [&](InsertPointTy IP) {
  if (IP.getBlock()->end() != IP.getPoint())
    return FiniCB(IP);
  // This must be done otherwise any nested constructs using FinalizeOMPRegion
  // will fail because that function requires the Finalization Basic Block to
  // have a terminator, which is already removed by EmitOMPRegionBody.
  // IP is currently at cancelation block.
  // We need to backtrack to the condition block to fetch
  // the exit block and create a branch from cancelation
  // to exit block.
  IRBuilder<>::InsertPointGuard IPG(Builder);
  Builder.restoreIP(IP);
  auto *CaseBB = IP.getBlock()->getSinglePredecessor();
  auto *CondBB = CaseBB->getSinglePredecessor()->getSinglePredecessor();
  auto *ExitBB = CondBB->getTerminator()->getSuccessor(1);
  Instruction *I = Builder.CreateBr(ExitBB);
  IP = InsertPointTy(I->getParent(), I->getIterator());
  return FiniCB(IP);
};

FinalizationStack.push_back({FiniCBWrapper, OMPD_sections, IsCancellable});

// Each section is emitted as a switch case
// Each finalization callback is handled from clang.EmitOMPSectionDirective()
// -> OMP.createSection() which generates the IR for each section
// Iterate through all sections and emit a switch construct:
// switch (IV) {
//   case 0:
//     <SectionStmt[0]>;
//     break;
// ...
//   case <NumSection> - 1:
//     <SectionStmt[<NumSection> - 1]>;
//     break;
// }
// ...
// section_loop.after:
// <FiniCB>;
auto LoopBodyGenCB = [&](InsertPointTy CodeGenIP, Value *IndVar) {
  Builder.restoreIP(CodeGenIP);
  BasicBlock *Continue =
      splitBBWithSuffix(Builder, /*CreateBranch=*/false, ".sections.after");
  Function *CurFn = Continue->getParent();
  SwitchInst *SwitchStmt = Builder.CreateSwitch(IndVar, Continue);

  unsigned CaseNumber = 0;
  for (auto SectionCB : SectionCBs) {
    BasicBlock *CaseBB = BasicBlock::Create(
        M.getContext(), "omp_section_loop.body.case", CurFn, Continue);
    SwitchStmt->addCase(Builder.getInt32(CaseNumber), CaseBB);
    Builder.SetInsertPoint(CaseBB);
    BranchInst *CaseEndBr = Builder.CreateBr(Continue);
    SectionCB(InsertPointTy(),
              {CaseEndBr->getParent(), CaseEndBr->getIterator()});
    CaseNumber++;
  }
  // remove the existing terminator from body BB since there can be no
  // terminators after switch/case
};
// Loop body ends here
// LowerBound, UpperBound, and STride for createCanonicalLoop
Type *I32Ty = Type::getInt32Ty(M.getContext());
Value *LB = ConstantInt::get(I32Ty, 0);
Value *UB = ConstantInt::get(I32Ty, SectionCBs.size());
Value *ST = ConstantInt::get(I32Ty, 1);
llvm::CanonicalLoopInfo *LoopInfo = createCanonicalLoop(
    Loc, LoopBodyGenCB, LB, UB, ST, true, false, AllocaIP, "section_loop");
InsertPointTy AfterIP =
    applyStaticWorkshareLoop(Loc.DL, LoopInfo, AllocaIP, !IsNowait);

// Apply the finalization callback in LoopAfterBB
auto FiniInfo = FinalizationStack.pop_back_val();
assert(FiniInfo.DK == OMPD_sections &&(static_cast <bool> (FiniInfo.DK == OMPD_sections &&
 "Unexpected finalization stack state!") ? void (0) : __assert_fail
 ("FiniInfo.DK == OMPD_sections && \"Unexpected finalization stack state!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1664, __extension__
 __PRETTY_FUNCTION__))
       "Unexpected finalization stack state!")(static_cast <bool> (FiniInfo.DK == OMPD_sections &&
 "Unexpected finalization stack state!") ? void (0) : __assert_fail
 ("FiniInfo.DK == OMPD_sections && \"Unexpected finalization stack state!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1664, __extension__
 __PRETTY_FUNCTION__));
if (FinalizeCallbackTy &CB = FiniInfo.FiniCB) {
  Builder.restoreIP(AfterIP);
  BasicBlock *FiniBB =
      splitBBWithSuffix(Builder, /*CreateBranch=*/true, "sections.fini");
  CB(Builder.saveIP());
  AfterIP = {FiniBB, FiniBB->begin()};
}

return AfterIP;
1674}

1676OpenMPIRBuilder::InsertPointTy
1677OpenMPIRBuilder::createSection(const LocationDescription &Loc,
                             BodyGenCallbackTy BodyGenCB,
                             FinalizeCallbackTy FiniCB) {
if (!updateToLocation(Loc))
  return Loc.IP;

auto FiniCBWrapper = [&](InsertPointTy IP) {
  if (IP.getBlock()->end() != IP.getPoint())
    return FiniCB(IP);
  // This must be done otherwise any nested constructs using FinalizeOMPRegion
  // will fail because that function requires the Finalization Basic Block to
  // have a terminator, which is already removed by EmitOMPRegionBody.
  // IP is currently at cancelation block.
  // We need to backtrack to the condition block to fetch
  // the exit block and create a branch from cancelation
  // to exit block.
  IRBuilder<>::InsertPointGuard IPG(Builder);
  Builder.restoreIP(IP);
  auto *CaseBB = Loc.IP.getBlock();
  auto *CondBB = CaseBB->getSinglePredecessor()->getSinglePredecessor();
  auto *ExitBB = CondBB->getTerminator()->getSuccessor(1);
  Instruction *I = Builder.CreateBr(ExitBB);
  IP = InsertPointTy(I->getParent(), I->getIterator());
  return FiniCB(IP);
};

Directive OMPD = Directive::OMPD_sections;
// Since we are using Finalization Callback here, HasFinalize
// and IsCancellable have to be true
return EmitOMPInlinedRegion(OMPD, nullptr, nullptr, BodyGenCB, FiniCBWrapper,
                            /*Conditional*/ false, /*hasFinalize*/ true,
                            /*IsCancellable*/ true);
1709}

1711/// Create a function with a unique name and a "void (i8*, i8*)" signature in
1712/// the given module and return it.
1713Function *getFreshReductionFunc(Module &M) {
Type *VoidTy = Type::getVoidTy(M.getContext());
Type *Int8PtrTy = Type::getInt8PtrTy(M.getContext());
auto *FuncTy =
    FunctionType::get(VoidTy, {Int8PtrTy, Int8PtrTy}, /* IsVarArg */ false);
return Function::Create(FuncTy, GlobalVariable::InternalLinkage,
                        M.getDataLayout().getDefaultGlobalsAddressSpace(),
                        ".omp.reduction.func", &M);
1721}

1723OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::createReductions(
  const LocationDescription &Loc, InsertPointTy AllocaIP,
  ArrayRef<ReductionInfo> ReductionInfos, bool IsNoWait) {
for (const ReductionInfo &RI : ReductionInfos) {
  (void)RI;
  assert(RI.Variable && "expected non-null variable")(static_cast <bool> (RI.Variable && "expected non-null variable"
) ? void (0) : __assert_fail ("RI.Variable && \"expected non-null variable\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1728, __extension__
 __PRETTY_FUNCTION__));
  assert(RI.PrivateVariable && "expected non-null private variable")(static_cast <bool> (RI.PrivateVariable && "expected non-null private variable"
) ? void (0) : __assert_fail ("RI.PrivateVariable && \"expected non-null private variable\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1729, __extension__
 __PRETTY_FUNCTION__));
  assert(RI.ReductionGen && "expected non-null reduction generator callback")(static_cast <bool> (RI.ReductionGen && "expected non-null reduction generator callback"
) ? void (0) : __assert_fail ("RI.ReductionGen && \"expected non-null reduction generator callback\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1730, __extension__
 __PRETTY_FUNCTION__));
  assert(RI.Variable->getType() == RI.PrivateVariable->getType() &&(static_cast <bool> (RI.Variable->getType() == RI.PrivateVariable
->getType() && "expected variables and their private equivalents to have the same "
 "type") ? void (0) : __assert_fail ("RI.Variable->getType() == RI.PrivateVariable->getType() && \"expected variables and their private equivalents to have the same \" \"type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1733, __extension__
 __PRETTY_FUNCTION__))
         "expected variables and their private equivalents to have the same "(static_cast <bool> (RI.Variable->getType() == RI.PrivateVariable
->getType() && "expected variables and their private equivalents to have the same "
 "type") ? void (0) : __assert_fail ("RI.Variable->getType() == RI.PrivateVariable->getType() && \"expected variables and their private equivalents to have the same \" \"type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1733, __extension__
 __PRETTY_FUNCTION__))
         "type")(static_cast <bool> (RI.Variable->getType() == RI.PrivateVariable
->getType() && "expected variables and their private equivalents to have the same "
 "type") ? void (0) : __assert_fail ("RI.Variable->getType() == RI.PrivateVariable->getType() && \"expected variables and their private equivalents to have the same \" \"type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1733, __extension__
 __PRETTY_FUNCTION__));
  assert(RI.Variable->getType()->isPointerTy() &&(static_cast <bool> (RI.Variable->getType()->isPointerTy
() && "expected variables to be pointers") ? void (0)
 : __assert_fail ("RI.Variable->getType()->isPointerTy() && \"expected variables to be pointers\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1735, __extension__
 __PRETTY_FUNCTION__))
         "expected variables to be pointers")(static_cast <bool> (RI.Variable->getType()->isPointerTy
() && "expected variables to be pointers") ? void (0)
 : __assert_fail ("RI.Variable->getType()->isPointerTy() && \"expected variables to be pointers\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 1735, __extension__
 __PRETTY_FUNCTION__));
}

if (!updateToLocation(Loc))
  return InsertPointTy();

BasicBlock *InsertBlock = Loc.IP.getBlock();
BasicBlock *ContinuationBlock =
    InsertBlock->splitBasicBlock(Loc.IP.getPoint(), "reduce.finalize");
InsertBlock->getTerminator()->eraseFromParent();

// Create and populate array of type-erased pointers to private reduction
// values.
unsigned NumReductions = ReductionInfos.size();
Type *RedArrayTy = ArrayType::get(Builder.getInt8PtrTy(), NumReductions);
Builder.restoreIP(AllocaIP);
Value *RedArray = Builder.CreateAlloca(RedArrayTy, nullptr, "red.array");

Builder.SetInsertPoint(InsertBlock, InsertBlock->end());

for (auto En : enumerate(ReductionInfos)) {
  unsigned Index = En.index();
  const ReductionInfo &RI = En.value();
  Value *RedArrayElemPtr = Builder.CreateConstInBoundsGEP2_64(
      RedArrayTy, RedArray, 0, Index, "red.array.elem." + Twine(Index));
  Value *Casted =
      Builder.CreateBitCast(RI.PrivateVariable, Builder.getInt8PtrTy(),
                            "private.red.var." + Twine(Index) + ".casted");
  Builder.CreateStore(Casted, RedArrayElemPtr);
}

// Emit a call to the runtime function that orchestrates the reduction.
// Declare the reduction function in the process.
Function *Func = Builder.GetInsertBlock()->getParent();
Module *Module = Func->getParent();
Value *RedArrayPtr =
    Builder.CreateBitCast(RedArray, Builder.getInt8PtrTy(), "red.array.ptr");
uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
bool CanGenerateAtomic =
    llvm::all_of(ReductionInfos, [](const ReductionInfo &RI) {
      return RI.AtomicReductionGen;
    });
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize,
                                CanGenerateAtomic
                                    ? IdentFlag::OMP_IDENT_FLAG_ATOMIC_REDUCE
                                    : IdentFlag(0));
Value *ThreadId = getOrCreateThreadID(Ident);
Constant *NumVariables = Builder.getInt32(NumReductions);
const DataLayout &DL = Module->getDataLayout();
unsigned RedArrayByteSize = DL.getTypeStoreSize(RedArrayTy);
Constant *RedArraySize = Builder.getInt64(RedArrayByteSize);
Function *ReductionFunc = getFreshReductionFunc(*Module);
Value *Lock = getOMPCriticalRegionLock(".reduction");
Function *ReduceFunc = getOrCreateRuntimeFunctionPtr(
    IsNoWait ? RuntimeFunction::OMPRTL___kmpc_reduce_nowait
             : RuntimeFunction::OMPRTL___kmpc_reduce);
CallInst *ReduceCall =
    Builder.CreateCall(ReduceFunc,
                       {Ident, ThreadId, NumVariables, RedArraySize,
                        RedArrayPtr, ReductionFunc, Lock},
                       "reduce");

// Create final reduction entry blocks for the atomic and non-atomic case.
// Emit IR that dispatches control flow to one of the blocks based on the
// reduction supporting the atomic mode.
BasicBlock *NonAtomicRedBlock =
    BasicBlock::Create(Module->getContext(), "reduce.switch.nonatomic", Func);
BasicBlock *AtomicRedBlock =
    BasicBlock::Create(Module->getContext(), "reduce.switch.atomic", Func);
SwitchInst *Switch =
    Builder.CreateSwitch(ReduceCall, ContinuationBlock, /* NumCases */ 2);
Switch->addCase(Builder.getInt32(1), NonAtomicRedBlock);
Switch->addCase(Builder.getInt32(2), AtomicRedBlock);

// Populate the non-atomic reduction using the elementwise reduction function.
// This loads the elements from the global and private variables and reduces
// them before storing back the result to the global variable.
Builder.SetInsertPoint(NonAtomicRedBlock);
for (auto En : enumerate(ReductionInfos)) {
  const ReductionInfo &RI = En.value();
  Type *ValueType = RI.ElementType;
  Value *RedValue = Builder.CreateLoad(ValueType, RI.Variable,
                                       "red.value." + Twine(En.index()));
  Value *PrivateRedValue =
      Builder.CreateLoad(ValueType, RI.PrivateVariable,
                         "red.private.value." + Twine(En.index()));
  Value *Reduced;
  Builder.restoreIP(
      RI.ReductionGen(Builder.saveIP(), RedValue, PrivateRedValue, Reduced));
  if (!Builder.GetInsertBlock())
    return InsertPointTy();
  Builder.CreateStore(Reduced, RI.Variable);
}
Function *EndReduceFunc = getOrCreateRuntimeFunctionPtr(
    IsNoWait ? RuntimeFunction::OMPRTL___kmpc_end_reduce_nowait
             : RuntimeFunction::OMPRTL___kmpc_end_reduce);
Builder.CreateCall(EndReduceFunc, {Ident, ThreadId, Lock});
Builder.CreateBr(ContinuationBlock);

// Populate the atomic reduction using the atomic elementwise reduction
// function. There are no loads/stores here because they will be happening
// inside the atomic elementwise reduction.
Builder.SetInsertPoint(AtomicRedBlock);
if (CanGenerateAtomic) {
  for (const ReductionInfo &RI : ReductionInfos) {
    Builder.restoreIP(RI.AtomicReductionGen(Builder.saveIP(), RI.ElementType,
                                            RI.Variable, RI.PrivateVariable));
    if (!Builder.GetInsertBlock())
      return InsertPointTy();
  }
  Builder.CreateBr(ContinuationBlock);
} else {
  Builder.CreateUnreachable();
}

// Populate the outlined reduction function using the elementwise reduction
// function. Partial values are extracted from the type-erased array of
// pointers to private variables.
BasicBlock *ReductionFuncBlock =
    BasicBlock::Create(Module->getContext(), "", ReductionFunc);
Builder.SetInsertPoint(ReductionFuncBlock);
Value *LHSArrayPtr = Builder.CreateBitCast(ReductionFunc->getArg(0),
                                           RedArrayTy->getPointerTo());
Value *RHSArrayPtr = Builder.CreateBitCast(ReductionFunc->getArg(1),
                                           RedArrayTy->getPointerTo());
for (auto En : enumerate(ReductionInfos)) {
  const ReductionInfo &RI = En.value();
  Value *LHSI8PtrPtr = Builder.CreateConstInBoundsGEP2_64(
      RedArrayTy, LHSArrayPtr, 0, En.index());
  Value *LHSI8Ptr = Builder.CreateLoad(Builder.getInt8PtrTy(), LHSI8PtrPtr);
  Value *LHSPtr = Builder.CreateBitCast(LHSI8Ptr, RI.Variable->getType());
  Value *LHS = Builder.CreateLoad(RI.ElementType, LHSPtr);
  Value *RHSI8PtrPtr = Builder.CreateConstInBoundsGEP2_64(
      RedArrayTy, RHSArrayPtr, 0, En.index());
  Value *RHSI8Ptr = Builder.CreateLoad(Builder.getInt8PtrTy(), RHSI8PtrPtr);
  Value *RHSPtr =
      Builder.CreateBitCast(RHSI8Ptr, RI.PrivateVariable->getType());
  Value *RHS = Builder.CreateLoad(RI.ElementType, RHSPtr);
  Value *Reduced;
  Builder.restoreIP(RI.ReductionGen(Builder.saveIP(), LHS, RHS, Reduced));
  if (!Builder.GetInsertBlock())
    return InsertPointTy();
  Builder.CreateStore(Reduced, LHSPtr);
}
Builder.CreateRetVoid();

Builder.SetInsertPoint(ContinuationBlock);
return Builder.saveIP();
1884}

1886OpenMPIRBuilder::InsertPointTy
1887OpenMPIRBuilder::createMaster(const LocationDescription &Loc,
                            BodyGenCallbackTy BodyGenCB,
                            FinalizeCallbackTy FiniCB) {

if (!updateToLocation(Loc))
  return Loc.IP;

Directive OMPD = Directive::OMPD_master;
uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
Value *ThreadId = getOrCreateThreadID(Ident);
Value *Args[] = {Ident, ThreadId};

Function *EntryRTLFn = getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_master);
Instruction *EntryCall = Builder.CreateCall(EntryRTLFn, Args);

Function *ExitRTLFn = getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_end_master);
Instruction *ExitCall = Builder.CreateCall(ExitRTLFn, Args);

return EmitOMPInlinedRegion(OMPD, EntryCall, ExitCall, BodyGenCB, FiniCB,
                            /*Conditional*/ true, /*hasFinalize*/ true);
1909}

1911OpenMPIRBuilder::InsertPointTy
1912OpenMPIRBuilder::createMasked(const LocationDescription &Loc,
                            BodyGenCallbackTy BodyGenCB,
                            FinalizeCallbackTy FiniCB, Value *Filter) {
if (!updateToLocation(Loc))
  return Loc.IP;

Directive OMPD = Directive::OMPD_masked;
uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
Value *ThreadId = getOrCreateThreadID(Ident);
Value *Args[] = {Ident, ThreadId, Filter};
Value *ArgsEnd[] = {Ident, ThreadId};

Function *EntryRTLFn = getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_masked);
Instruction *EntryCall = Builder.CreateCall(EntryRTLFn, Args);

Function *ExitRTLFn = getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_end_masked);
Instruction *ExitCall = Builder.CreateCall(ExitRTLFn, ArgsEnd);

return EmitOMPInlinedRegion(OMPD, EntryCall, ExitCall, BodyGenCB, FiniCB,
                            /*Conditional*/ true, /*hasFinalize*/ true);
1934}

1936CanonicalLoopInfo *OpenMPIRBuilder::createLoopSkeleton(
  DebugLoc DL, Value *TripCount, Function *F, BasicBlock *PreInsertBefore,
  BasicBlock *PostInsertBefore, const Twine &Name) {
Module *M = F->getParent();
LLVMContext &Ctx = M->getContext();
Type *IndVarTy = TripCount->getType();
11
←
Called C++ object pointer is null

// Create the basic block structure.
BasicBlock *Preheader =
    BasicBlock::Create(Ctx, "omp_" + Name + ".preheader", F, PreInsertBefore);
BasicBlock *Header =
    BasicBlock::Create(Ctx, "omp_" + Name + ".header", F, PreInsertBefore);
BasicBlock *Cond =
    BasicBlock::Create(Ctx, "omp_" + Name + ".cond", F, PreInsertBefore);
BasicBlock *Body =
    BasicBlock::Create(Ctx, "omp_" + Name + ".body", F, PreInsertBefore);
BasicBlock *Latch =
    BasicBlock::Create(Ctx, "omp_" + Name + ".inc", F, PostInsertBefore);
BasicBlock *Exit =
    BasicBlock::Create(Ctx, "omp_" + Name + ".exit", F, PostInsertBefore);
BasicBlock *After =
    BasicBlock::Create(Ctx, "omp_" + Name + ".after", F, PostInsertBefore);

// Use specified DebugLoc for new instructions.
Builder.SetCurrentDebugLocation(DL);

Builder.SetInsertPoint(Preheader);
Builder.CreateBr(Header);

Builder.SetInsertPoint(Header);
PHINode *IndVarPHI = Builder.CreatePHI(IndVarTy, 2, "omp_" + Name + ".iv");
IndVarPHI->addIncoming(ConstantInt::get(IndVarTy, 0), Preheader);
Builder.CreateBr(Cond);

Builder.SetInsertPoint(Cond);
Value *Cmp =
    Builder.CreateICmpULT(IndVarPHI, TripCount, "omp_" + Name + ".cmp");
Builder.CreateCondBr(Cmp, Body, Exit);

Builder.SetInsertPoint(Body);
Builder.CreateBr(Latch);

Builder.SetInsertPoint(Latch);
Value *Next = Builder.CreateAdd(IndVarPHI, ConstantInt::get(IndVarTy, 1),
                                "omp_" + Name + ".next", /*HasNUW=*/true);
Builder.CreateBr(Header);
IndVarPHI->addIncoming(Next, Latch);

Builder.SetInsertPoint(Exit);
Builder.CreateBr(After);

// Remember and return the canonical control flow.
LoopInfos.emplace_front();
CanonicalLoopInfo *CL = &LoopInfos.front();

CL->Header = Header;
CL->Cond = Cond;
CL->Latch = Latch;
CL->Exit = Exit;

1996#ifndef NDEBUG
CL->assertOK();
1998#endif
return CL;
2000}

2002CanonicalLoopInfo *
2003OpenMPIRBuilder::createCanonicalLoop(const LocationDescription &Loc,
                                   LoopBodyGenCallbackTy BodyGenCB,
                                   Value *TripCount, const Twine &Name) {
BasicBlock *BB = Loc.IP.getBlock();
BasicBlock *NextBB = BB->getNextNode();

CanonicalLoopInfo *CL = createLoopSkeleton(Loc.DL, TripCount, BB->getParent(),
                                           NextBB, NextBB, Name);
BasicBlock *After = CL->getAfter();

// If location is not set, don't connect the loop.
if (updateToLocation(Loc)) {
  // Split the loop at the insertion point: Branch to the preheader and move
  // every following instruction to after the loop (the After BB). Also, the
  // new successor is the loop's after block.
  spliceBB(Builder, After, /*CreateBranch=*/false);
  Builder.CreateBr(CL->getPreheader());
}

// Emit the body content. We do it after connecting the loop to the CFG to
// avoid that the callback encounters degenerate BBs.
BodyGenCB(CL->getBodyIP(), CL->getIndVar());

2026#ifndef NDEBUG
CL->assertOK();
2028#endif
return CL;
2030}

2032CanonicalLoopInfo *OpenMPIRBuilder::createCanonicalLoop(
  const LocationDescription &Loc, LoopBodyGenCallbackTy BodyGenCB,
  Value *Start, Value *Stop, Value *Step, bool IsSigned, bool InclusiveStop,
  InsertPointTy ComputeIP, const Twine &Name) {

// Consider the following difficulties (assuming 8-bit signed integers):
//  * Adding \p Step to the loop counter which passes \p Stop may overflow:
//      DO I = 1, 100, 50
///  * A \p Step of INT_MIN cannot not be normalized to a positive direction:
//      DO I = 100, 0, -128

// Start, Stop and Step must be of the same integer type.
auto *IndVarTy = cast<IntegerType>(Start->getType());
assert(IndVarTy == Stop->getType() && "Stop type mismatch")(static_cast <bool> (IndVarTy == Stop->getType() &&
 "Stop type mismatch") ? void (0) : __assert_fail ("IndVarTy == Stop->getType() && \"Stop type mismatch\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2045, __extension__
 __PRETTY_FUNCTION__));
assert(IndVarTy == Step->getType() && "Step type mismatch")(static_cast <bool> (IndVarTy == Step->getType() &&
 "Step type mismatch") ? void (0) : __assert_fail ("IndVarTy == Step->getType() && \"Step type mismatch\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2046, __extension__
 __PRETTY_FUNCTION__));

LocationDescription ComputeLoc =
    ComputeIP.isSet() ? LocationDescription(ComputeIP, Loc.DL) : Loc;
updateToLocation(ComputeLoc);

ConstantInt *Zero = ConstantInt::get(IndVarTy, 0);
ConstantInt *One = ConstantInt::get(IndVarTy, 1);

// Like Step, but always positive.
Value *Incr = Step;

// Distance between Start and Stop; always positive.
Value *Span;

// Condition whether there are no iterations are executed at all, e.g. because
// UB < LB.
Value *ZeroCmp;

if (IsSigned) {
  // Ensure that increment is positive. If not, negate and invert LB and UB.
  Value *IsNeg = Builder.CreateICmpSLT(Step, Zero);
  Incr = Builder.CreateSelect(IsNeg, Builder.CreateNeg(Step), Step);
  Value *LB = Builder.CreateSelect(IsNeg, Stop, Start);
  Value *UB = Builder.CreateSelect(IsNeg, Start, Stop);
  Span = Builder.CreateSub(UB, LB, "", false, true);
  ZeroCmp = Builder.CreateICmp(
      InclusiveStop ? CmpInst::ICMP_SLT : CmpInst::ICMP_SLE, UB, LB);
} else {
  Span = Builder.CreateSub(Stop, Start, "", true);
  ZeroCmp = Builder.CreateICmp(
      InclusiveStop ? CmpInst::ICMP_ULT : CmpInst::ICMP_ULE, Stop, Start);
}

Value *CountIfLooping;
if (InclusiveStop) {
  CountIfLooping = Builder.CreateAdd(Builder.CreateUDiv(Span, Incr), One);
} else {
  // Avoid incrementing past stop since it could overflow.
  Value *CountIfTwo = Builder.CreateAdd(
      Builder.CreateUDiv(Builder.CreateSub(Span, One), Incr), One);
  Value *OneCmp = Builder.CreateICmp(
      InclusiveStop ? CmpInst::ICMP_ULT : CmpInst::ICMP_ULE, Span, Incr);
  CountIfLooping = Builder.CreateSelect(OneCmp, One, CountIfTwo);
}
Value *TripCount = Builder.CreateSelect(ZeroCmp, Zero, CountIfLooping,
                                        "omp_" + Name + ".tripcount");

auto BodyGen = [=](InsertPointTy CodeGenIP, Value *IV) {
  Builder.restoreIP(CodeGenIP);
  Value *Span = Builder.CreateMul(IV, Step);
  Value *IndVar = Builder.CreateAdd(Span, Start);
  BodyGenCB(Builder.saveIP(), IndVar);
};
LocationDescription LoopLoc = ComputeIP.isSet() ? Loc.IP : Builder.saveIP();
return createCanonicalLoop(LoopLoc, BodyGen, TripCount, Name);
2102}

2104// Returns an LLVM function to call for initializing loop bounds using OpenMP
2105// static scheduling depending on `type`. Only i32 and i64 are supported by the
2106// runtime. Always interpret integers as unsigned similarly to
2107// CanonicalLoopInfo.
2108static FunctionCallee getKmpcForStaticInitForType(Type *Ty, Module &M,
                                                OpenMPIRBuilder &OMPBuilder) {
unsigned Bitwidth = Ty->getIntegerBitWidth();
if (Bitwidth == 32)
  return OMPBuilder.getOrCreateRuntimeFunction(
      M, omp::RuntimeFunction::OMPRTL___kmpc_for_static_init_4u);
if (Bitwidth == 64)
  return OMPBuilder.getOrCreateRuntimeFunction(
      M, omp::RuntimeFunction::OMPRTL___kmpc_for_static_init_8u);
llvm_unreachable("unknown OpenMP loop iterator bitwidth")::llvm::llvm_unreachable_internal("unknown OpenMP loop iterator bitwidth"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2117);
2118}

2120OpenMPIRBuilder::InsertPointTy
2121OpenMPIRBuilder::applyStaticWorkshareLoop(DebugLoc DL, CanonicalLoopInfo *CLI,
                                        InsertPointTy AllocaIP,
                                        bool NeedsBarrier) {
assert(CLI->isValid() && "Requires a valid canonical loop")(static_cast <bool> (CLI->isValid() && "Requires a valid canonical loop"
) ? void (0) : __assert_fail ("CLI->isValid() && \"Requires a valid canonical loop\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2124, __extension__
 __PRETTY_FUNCTION__));
assert(!isConflictIP(AllocaIP, CLI->getPreheaderIP()) &&(static_cast <bool> (!isConflictIP(AllocaIP, CLI->getPreheaderIP
()) && "Require dedicated allocate IP") ? void (0) : __assert_fail
 ("!isConflictIP(AllocaIP, CLI->getPreheaderIP()) && \"Require dedicated allocate IP\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2126, __extension__
 __PRETTY_FUNCTION__))
       "Require dedicated allocate IP")(static_cast <bool> (!isConflictIP(AllocaIP, CLI->getPreheaderIP
()) && "Require dedicated allocate IP") ? void (0) : __assert_fail
 ("!isConflictIP(AllocaIP, CLI->getPreheaderIP()) && \"Require dedicated allocate IP\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2126, __extension__
 __PRETTY_FUNCTION__));

// Set up the source location value for OpenMP runtime.
Builder.restoreIP(CLI->getPreheaderIP());
Builder.SetCurrentDebugLocation(DL);

uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(DL, SrcLocStrSize);
Value *SrcLoc = getOrCreateIdent(SrcLocStr, SrcLocStrSize);

// Declare useful OpenMP runtime functions.
Value *IV = CLI->getIndVar();
Type *IVTy = IV->getType();
FunctionCallee StaticInit = getKmpcForStaticInitForType(IVTy, M, *this);
FunctionCallee StaticFini =
    getOrCreateRuntimeFunction(M, omp::OMPRTL___kmpc_for_static_fini);

// Allocate space for computed loop bounds as expected by the "init" function.
Builder.restoreIP(AllocaIP);
Type *I32Type = Type::getInt32Ty(M.getContext());
Value *PLastIter = Builder.CreateAlloca(I32Type, nullptr, "p.lastiter");
Value *PLowerBound = Builder.CreateAlloca(IVTy, nullptr, "p.lowerbound");
Value *PUpperBound = Builder.CreateAlloca(IVTy, nullptr, "p.upperbound");
Value *PStride = Builder.CreateAlloca(IVTy, nullptr, "p.stride");

// At the end of the preheader, prepare for calling the "init" function by
// storing the current loop bounds into the allocated space. A canonical loop
// always iterates from 0 to trip-count with step 1. Note that "init" expects
// and produces an inclusive upper bound.
Builder.SetInsertPoint(CLI->getPreheader()->getTerminator());
Constant *Zero = ConstantInt::get(IVTy, 0);
Constant *One = ConstantInt::get(IVTy, 1);
Builder.CreateStore(Zero, PLowerBound);
Value *UpperBound = Builder.CreateSub(CLI->getTripCount(), One);
Builder.CreateStore(UpperBound, PUpperBound);
Builder.CreateStore(One, PStride);

Value *ThreadNum = getOrCreateThreadID(SrcLoc);

Constant *SchedulingType = ConstantInt::get(
    I32Type, static_cast<int>(OMPScheduleType::UnorderedStatic));

// Call the "init" function and update the trip count of the loop with the
// value it produced.
Builder.CreateCall(StaticInit,
                   {SrcLoc, ThreadNum, SchedulingType, PLastIter, PLowerBound,
                    PUpperBound, PStride, One, Zero});
Value *LowerBound = Builder.CreateLoad(IVTy, PLowerBound);
Value *InclusiveUpperBound = Builder.CreateLoad(IVTy, PUpperBound);
Value *TripCountMinusOne = Builder.CreateSub(InclusiveUpperBound, LowerBound);
Value *TripCount = Builder.CreateAdd(TripCountMinusOne, One);
CLI->setTripCount(TripCount);

// Update all uses of the induction variable except the one in the condition
// block that compares it with the actual upper bound, and the increment in
// the latch block.

CLI->mapIndVar([&](Instruction *OldIV) -> Value * {
  Builder.SetInsertPoint(CLI->getBody(),
                         CLI->getBody()->getFirstInsertionPt());
  Builder.SetCurrentDebugLocation(DL);
  return Builder.CreateAdd(OldIV, LowerBound);
});

// In the "exit" block, call the "fini" function.
Builder.SetInsertPoint(CLI->getExit(),
                       CLI->getExit()->getTerminator()->getIterator());
Builder.CreateCall(StaticFini, {SrcLoc, ThreadNum});

// Add the barrier if requested.
if (NeedsBarrier)
  createBarrier(LocationDescription(Builder.saveIP(), DL),
                omp::Directive::OMPD_for, /* ForceSimpleCall */ false,
                /* CheckCancelFlag */ false);

InsertPointTy AfterIP = CLI->getAfterIP();
CLI->invalidate();

return AfterIP;
2205}

2207OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::applyStaticChunkedWorkshareLoop(
  DebugLoc DL, CanonicalLoopInfo *CLI, InsertPointTy AllocaIP,
  bool NeedsBarrier, Value *ChunkSize) {
assert(CLI->isValid() && "Requires a valid canonical loop")(static_cast <bool> (CLI->isValid() && "Requires a valid canonical loop"
) ? void (0) : __assert_fail ("CLI->isValid() && \"Requires a valid canonical loop\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2210, __extension__
 __PRETTY_FUNCTION__));
assert(ChunkSize && "Chunk size is required")(static_cast <bool> (ChunkSize && "Chunk size is required"
) ? void (0) : __assert_fail ("ChunkSize && \"Chunk size is required\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2211, __extension__
 __PRETTY_FUNCTION__));

LLVMContext &Ctx = CLI->getFunction()->getContext();
Value *IV = CLI->getIndVar();
Value *OrigTripCount = CLI->getTripCount();
Type *IVTy = IV->getType();
assert(IVTy->getIntegerBitWidth() <= 64 &&(static_cast <bool> (IVTy->getIntegerBitWidth() <=
&& "Max supported tripcount bitwidth is 64 bits")
 ? void (0) : __assert_fail ("IVTy->getIntegerBitWidth() <= 64 && \"Max supported tripcount bitwidth is 64 bits\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2218, __extension__
 __PRETTY_FUNCTION__))
       "Max supported tripcount bitwidth is 64 bits")(static_cast <bool> (IVTy->getIntegerBitWidth() <=
&& "Max supported tripcount bitwidth is 64 bits")
 ? void (0) : __assert_fail ("IVTy->getIntegerBitWidth() <= 64 && \"Max supported tripcount bitwidth is 64 bits\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2218, __extension__
 __PRETTY_FUNCTION__));
Type *InternalIVTy = IVTy->getIntegerBitWidth() <= 32 ? Type::getInt32Ty(Ctx)
                                                      : Type::getInt64Ty(Ctx);
Type *I32Type = Type::getInt32Ty(M.getContext());
Constant *Zero = ConstantInt::get(InternalIVTy, 0);
Constant *One = ConstantInt::get(InternalIVTy, 1);

// Declare useful OpenMP runtime functions.
FunctionCallee StaticInit =
    getKmpcForStaticInitForType(InternalIVTy, M, *this);
FunctionCallee StaticFini =
    getOrCreateRuntimeFunction(M, omp::OMPRTL___kmpc_for_static_fini);

// Allocate space for computed loop bounds as expected by the "init" function.
Builder.restoreIP(AllocaIP);
Builder.SetCurrentDebugLocation(DL);
Value *PLastIter = Builder.CreateAlloca(I32Type, nullptr, "p.lastiter");
Value *PLowerBound =
    Builder.CreateAlloca(InternalIVTy, nullptr, "p.lowerbound");
Value *PUpperBound =
    Builder.CreateAlloca(InternalIVTy, nullptr, "p.upperbound");
Value *PStride = Builder.CreateAlloca(InternalIVTy, nullptr, "p.stride");

// Set up the source location value for the OpenMP runtime.
Builder.restoreIP(CLI->getPreheaderIP());
Builder.SetCurrentDebugLocation(DL);

// TODO: Detect overflow in ubsan or max-out with current tripcount.
Value *CastedChunkSize =
    Builder.CreateZExtOrTrunc(ChunkSize, InternalIVTy, "chunksize");
Value *CastedTripCount =
    Builder.CreateZExt(OrigTripCount, InternalIVTy, "tripcount");

Constant *SchedulingType = ConstantInt::get(
    I32Type, static_cast<int>(OMPScheduleType::UnorderedStaticChunked));
Builder.CreateStore(Zero, PLowerBound);
Value *OrigUpperBound = Builder.CreateSub(CastedTripCount, One);
Builder.CreateStore(OrigUpperBound, PUpperBound);
Builder.CreateStore(One, PStride);

// Call the "init" function and update the trip count of the loop with the
// value it produced.
uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(DL, SrcLocStrSize);
Value *SrcLoc = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
Value *ThreadNum = getOrCreateThreadID(SrcLoc);
Builder.CreateCall(StaticInit,
                   {/*loc=*/SrcLoc, /*global_tid=*/ThreadNum,
                    /*schedtype=*/SchedulingType, /*plastiter=*/PLastIter,
                    /*plower=*/PLowerBound, /*pupper=*/PUpperBound,
                    /*pstride=*/PStride, /*incr=*/One,
                    /*chunk=*/CastedChunkSize});

// Load values written by the "init" function.
Value *FirstChunkStart =
    Builder.CreateLoad(InternalIVTy, PLowerBound, "omp_firstchunk.lb");
Value *FirstChunkStop =
    Builder.CreateLoad(InternalIVTy, PUpperBound, "omp_firstchunk.ub");
Value *FirstChunkEnd = Builder.CreateAdd(FirstChunkStop, One);
Value *ChunkRange =
    Builder.CreateSub(FirstChunkEnd, FirstChunkStart, "omp_chunk.range");
Value *NextChunkStride =
    Builder.CreateLoad(InternalIVTy, PStride, "omp_dispatch.stride");

// Create outer "dispatch" loop for enumerating the chunks.
BasicBlock *DispatchEnter = splitBB(Builder, true);
Value *DispatchCounter;
CanonicalLoopInfo *DispatchCLI = createCanonicalLoop(
    {Builder.saveIP(), DL},
    [&](InsertPointTy BodyIP, Value *Counter) { DispatchCounter = Counter; },
    FirstChunkStart, CastedTripCount, NextChunkStride,
    /*IsSigned=*/false, /*InclusiveStop=*/false, /*ComputeIP=*/{},
    "dispatch");

// Remember the BasicBlocks of the dispatch loop we need, then invalidate to
// not have to preserve the canonical invariant.
BasicBlock *DispatchBody = DispatchCLI->getBody();
BasicBlock *DispatchLatch = DispatchCLI->getLatch();
BasicBlock *DispatchExit = DispatchCLI->getExit();
BasicBlock *DispatchAfter = DispatchCLI->getAfter();
DispatchCLI->invalidate();

// Rewire the original loop to become the chunk loop inside the dispatch loop.
redirectTo(DispatchAfter, CLI->getAfter(), DL);
redirectTo(CLI->getExit(), DispatchLatch, DL);
redirectTo(DispatchBody, DispatchEnter, DL);

// Prepare the prolog of the chunk loop.
Builder.restoreIP(CLI->getPreheaderIP());
Builder.SetCurrentDebugLocation(DL);

// Compute the number of iterations of the chunk loop.
Builder.SetInsertPoint(CLI->getPreheader()->getTerminator());
Value *ChunkEnd = Builder.CreateAdd(DispatchCounter, ChunkRange);
Value *IsLastChunk =
    Builder.CreateICmpUGE(ChunkEnd, CastedTripCount, "omp_chunk.is_last");
Value *CountUntilOrigTripCount =
    Builder.CreateSub(CastedTripCount, DispatchCounter);
Value *ChunkTripCount = Builder.CreateSelect(
    IsLastChunk, CountUntilOrigTripCount, ChunkRange, "omp_chunk.tripcount");
Value *BackcastedChunkTC =
    Builder.CreateTrunc(ChunkTripCount, IVTy, "omp_chunk.tripcount.trunc");
CLI->setTripCount(BackcastedChunkTC);

// Update all uses of the induction variable except the one in the condition
// block that compares it with the actual upper bound, and the increment in
// the latch block.
Value *BackcastedDispatchCounter =
    Builder.CreateTrunc(DispatchCounter, IVTy, "omp_dispatch.iv.trunc");
CLI->mapIndVar([&](Instruction *) -> Value * {
  Builder.restoreIP(CLI->getBodyIP());
  return Builder.CreateAdd(IV, BackcastedDispatchCounter);
});

// In the "exit" block, call the "fini" function.
Builder.SetInsertPoint(DispatchExit, DispatchExit->getFirstInsertionPt());
Builder.CreateCall(StaticFini, {SrcLoc, ThreadNum});

// Add the barrier if requested.
if (NeedsBarrier)
  createBarrier(LocationDescription(Builder.saveIP(), DL), OMPD_for,
                /*ForceSimpleCall=*/false, /*CheckCancelFlag=*/false);

2341#ifndef NDEBUG
// Even though we currently do not support applying additional methods to it,
// the chunk loop should remain a canonical loop.
CLI->assertOK();
2345#endif

return {DispatchAfter, DispatchAfter->getFirstInsertionPt()};
2348}

2350OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::applyWorkshareLoop(
  DebugLoc DL, CanonicalLoopInfo *CLI, InsertPointTy AllocaIP,
  bool NeedsBarrier, llvm::omp::ScheduleKind SchedKind,
  llvm::Value *ChunkSize, bool HasSimdModifier, bool HasMonotonicModifier,
  bool HasNonmonotonicModifier, bool HasOrderedClause) {
OMPScheduleType EffectiveScheduleType = computeOpenMPScheduleType(
    SchedKind, ChunkSize, HasSimdModifier, HasMonotonicModifier,
    HasNonmonotonicModifier, HasOrderedClause);

bool IsOrdered = (EffectiveScheduleType & OMPScheduleType::ModifierOrdered) ==
                 OMPScheduleType::ModifierOrdered;
switch (EffectiveScheduleType & ~OMPScheduleType::ModifierMask) {
case OMPScheduleType::BaseStatic:
  assert(!ChunkSize && "No chunk size with static-chunked schedule")(static_cast <bool> (!ChunkSize && "No chunk size with static-chunked schedule"
) ? void (0) : __assert_fail ("!ChunkSize && \"No chunk size with static-chunked schedule\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2363, __extension__
 __PRETTY_FUNCTION__));
  if (IsOrdered)
    return applyDynamicWorkshareLoop(DL, CLI, AllocaIP, EffectiveScheduleType,
                                     NeedsBarrier, ChunkSize);
  // FIXME: Monotonicity ignored?
  return applyStaticWorkshareLoop(DL, CLI, AllocaIP, NeedsBarrier);

case OMPScheduleType::BaseStaticChunked:
  if (IsOrdered)
    return applyDynamicWorkshareLoop(DL, CLI, AllocaIP, EffectiveScheduleType,
                                     NeedsBarrier, ChunkSize);
  // FIXME: Monotonicity ignored?
  return applyStaticChunkedWorkshareLoop(DL, CLI, AllocaIP, NeedsBarrier,
                                         ChunkSize);

case OMPScheduleType::BaseRuntime:
case OMPScheduleType::BaseAuto:
case OMPScheduleType::BaseGreedy:
case OMPScheduleType::BaseBalanced:
case OMPScheduleType::BaseSteal:
case OMPScheduleType::BaseGuidedSimd:
case OMPScheduleType::BaseRuntimeSimd:
  assert(!ChunkSize &&(static_cast <bool> (!ChunkSize && "schedule type does not support user-defined chunk sizes"
) ? void (0) : __assert_fail ("!ChunkSize && \"schedule type does not support user-defined chunk sizes\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2386, __extension__
 __PRETTY_FUNCTION__))
         "schedule type does not support user-defined chunk sizes")(static_cast <bool> (!ChunkSize && "schedule type does not support user-defined chunk sizes"
) ? void (0) : __assert_fail ("!ChunkSize && \"schedule type does not support user-defined chunk sizes\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2386, __extension__
 __PRETTY_FUNCTION__));
  LLVM_FALLTHROUGH[[fallthrough]];
case OMPScheduleType::BaseDynamicChunked:
case OMPScheduleType::BaseGuidedChunked:
case OMPScheduleType::BaseGuidedIterativeChunked:
case OMPScheduleType::BaseGuidedAnalyticalChunked:
case OMPScheduleType::BaseStaticBalancedChunked:
  return applyDynamicWorkshareLoop(DL, CLI, AllocaIP, EffectiveScheduleType,
                                   NeedsBarrier, ChunkSize);

default:
  llvm_unreachable("Unknown/unimplemented schedule kind")::llvm::llvm_unreachable_internal("Unknown/unimplemented schedule kind"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2397);
}
2399}

2401/// Returns an LLVM function to call for initializing loop bounds using OpenMP
2402/// dynamic scheduling depending on `type`. Only i32 and i64 are supported by
2403/// the runtime. Always interpret integers as unsigned similarly to
2404/// CanonicalLoopInfo.
2405static FunctionCallee
2406getKmpcForDynamicInitForType(Type *Ty, Module &M, OpenMPIRBuilder &OMPBuilder) {
unsigned Bitwidth = Ty->getIntegerBitWidth();
if (Bitwidth == 32)
  return OMPBuilder.getOrCreateRuntimeFunction(
      M, omp::RuntimeFunction::OMPRTL___kmpc_dispatch_init_4u);
if (Bitwidth == 64)
  return OMPBuilder.getOrCreateRuntimeFunction(
      M, omp::RuntimeFunction::OMPRTL___kmpc_dispatch_init_8u);
llvm_unreachable("unknown OpenMP loop iterator bitwidth")::llvm::llvm_unreachable_internal("unknown OpenMP loop iterator bitwidth"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2414);
2415}

2417/// Returns an LLVM function to call for updating the next loop using OpenMP
2418/// dynamic scheduling depending on `type`. Only i32 and i64 are supported by
2419/// the runtime. Always interpret integers as unsigned similarly to
2420/// CanonicalLoopInfo.
2421static FunctionCallee
2422getKmpcForDynamicNextForType(Type *Ty, Module &M, OpenMPIRBuilder &OMPBuilder) {
unsigned Bitwidth = Ty->getIntegerBitWidth();
if (Bitwidth == 32)
  return OMPBuilder.getOrCreateRuntimeFunction(
      M, omp::RuntimeFunction::OMPRTL___kmpc_dispatch_next_4u);
if (Bitwidth == 64)
  return OMPBuilder.getOrCreateRuntimeFunction(
      M, omp::RuntimeFunction::OMPRTL___kmpc_dispatch_next_8u);
llvm_unreachable("unknown OpenMP loop iterator bitwidth")::llvm::llvm_unreachable_internal("unknown OpenMP loop iterator bitwidth"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2430);
2431}

2433/// Returns an LLVM function to call for finalizing the dynamic loop using
2434/// depending on `type`. Only i32 and i64 are supported by the runtime. Always
2435/// interpret integers as unsigned similarly to CanonicalLoopInfo.
2436static FunctionCallee
2437getKmpcForDynamicFiniForType(Type *Ty, Module &M, OpenMPIRBuilder &OMPBuilder) {
unsigned Bitwidth = Ty->getIntegerBitWidth();
if (Bitwidth == 32)
  return OMPBuilder.getOrCreateRuntimeFunction(
      M, omp::RuntimeFunction::OMPRTL___kmpc_dispatch_fini_4u);
if (Bitwidth == 64)
  return OMPBuilder.getOrCreateRuntimeFunction(
      M, omp::RuntimeFunction::OMPRTL___kmpc_dispatch_fini_8u);
llvm_unreachable("unknown OpenMP loop iterator bitwidth")::llvm::llvm_unreachable_internal("unknown OpenMP loop iterator bitwidth"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2445);
2446}

2448OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::applyDynamicWorkshareLoop(
  DebugLoc DL, CanonicalLoopInfo *CLI, InsertPointTy AllocaIP,
  OMPScheduleType SchedType, bool NeedsBarrier, Value *Chunk) {
assert(CLI->isValid() && "Requires a valid canonical loop")(static_cast <bool> (CLI->isValid() && "Requires a valid canonical loop"
) ? void (0) : __assert_fail ("CLI->isValid() && \"Requires a valid canonical loop\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2451, __extension__
 __PRETTY_FUNCTION__));
assert(!isConflictIP(AllocaIP, CLI->getPreheaderIP()) &&(static_cast <bool> (!isConflictIP(AllocaIP, CLI->getPreheaderIP
()) && "Require dedicated allocate IP") ? void (0) : __assert_fail
 ("!isConflictIP(AllocaIP, CLI->getPreheaderIP()) && \"Require dedicated allocate IP\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2453, __extension__
 __PRETTY_FUNCTION__))
       "Require dedicated allocate IP")(static_cast <bool> (!isConflictIP(AllocaIP, CLI->getPreheaderIP
()) && "Require dedicated allocate IP") ? void (0) : __assert_fail
 ("!isConflictIP(AllocaIP, CLI->getPreheaderIP()) && \"Require dedicated allocate IP\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2453, __extension__
 __PRETTY_FUNCTION__));
assert(isValidWorkshareLoopScheduleType(SchedType) &&(static_cast <bool> (isValidWorkshareLoopScheduleType(SchedType
) && "Require valid schedule type") ? void (0) : __assert_fail
 ("isValidWorkshareLoopScheduleType(SchedType) && \"Require valid schedule type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2455, __extension__
 __PRETTY_FUNCTION__))
       "Require valid schedule type")(static_cast <bool> (isValidWorkshareLoopScheduleType(SchedType
) && "Require valid schedule type") ? void (0) : __assert_fail
 ("isValidWorkshareLoopScheduleType(SchedType) && \"Require valid schedule type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2455, __extension__
 __PRETTY_FUNCTION__));

bool Ordered = (SchedType & OMPScheduleType::ModifierOrdered) ==
               OMPScheduleType::ModifierOrdered;

// Set up the source location value for OpenMP runtime.
Builder.SetCurrentDebugLocation(DL);

uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(DL, SrcLocStrSize);
Value *SrcLoc = getOrCreateIdent(SrcLocStr, SrcLocStrSize);

// Declare useful OpenMP runtime functions.
Value *IV = CLI->getIndVar();
Type *IVTy = IV->getType();
FunctionCallee DynamicInit = getKmpcForDynamicInitForType(IVTy, M, *this);
FunctionCallee DynamicNext = getKmpcForDynamicNextForType(IVTy, M, *this);

// Allocate space for computed loop bounds as expected by the "init" function.
Builder.restoreIP(AllocaIP);
Type *I32Type = Type::getInt32Ty(M.getContext());
Value *PLastIter = Builder.CreateAlloca(I32Type, nullptr, "p.lastiter");
Value *PLowerBound = Builder.CreateAlloca(IVTy, nullptr, "p.lowerbound");
Value *PUpperBound = Builder.CreateAlloca(IVTy, nullptr, "p.upperbound");
Value *PStride = Builder.CreateAlloca(IVTy, nullptr, "p.stride");

// At the end of the preheader, prepare for calling the "init" function by
// storing the current loop bounds into the allocated space. A canonical loop
// always iterates from 0 to trip-count with step 1. Note that "init" expects
// and produces an inclusive upper bound.
BasicBlock *PreHeader = CLI->getPreheader();
Builder.SetInsertPoint(PreHeader->getTerminator());
Constant *One = ConstantInt::get(IVTy, 1);
Builder.CreateStore(One, PLowerBound);
Value *UpperBound = CLI->getTripCount();
Builder.CreateStore(UpperBound, PUpperBound);
Builder.CreateStore(One, PStride);

BasicBlock *Header = CLI->getHeader();
BasicBlock *Exit = CLI->getExit();
BasicBlock *Cond = CLI->getCond();
BasicBlock *Latch = CLI->getLatch();
InsertPointTy AfterIP = CLI->getAfterIP();

// The CLI will be "broken" in the code below, as the loop is no longer
// a valid canonical loop.

if (!Chunk)
  Chunk = One;

Value *ThreadNum = getOrCreateThreadID(SrcLoc);

Constant *SchedulingType =
    ConstantInt::get(I32Type, static_cast<int>(SchedType));

// Call the "init" function.
Builder.CreateCall(DynamicInit,
                   {SrcLoc, ThreadNum, SchedulingType, /* LowerBound */ One,
                    UpperBound, /* step */ One, Chunk});

// An outer loop around the existing one.
BasicBlock *OuterCond = BasicBlock::Create(
    PreHeader->getContext(), Twine(PreHeader->getName()) + ".outer.cond",
    PreHeader->getParent());
// This needs to be 32-bit always, so can't use the IVTy Zero above.
Builder.SetInsertPoint(OuterCond, OuterCond->getFirstInsertionPt());
Value *Res =
    Builder.CreateCall(DynamicNext, {SrcLoc, ThreadNum, PLastIter,
                                     PLowerBound, PUpperBound, PStride});
Constant *Zero32 = ConstantInt::get(I32Type, 0);
Value *MoreWork = Builder.CreateCmp(CmpInst::ICMP_NE, Res, Zero32);
Value *LowerBound =
    Builder.CreateSub(Builder.CreateLoad(IVTy, PLowerBound), One, "lb");
Builder.CreateCondBr(MoreWork, Header, Exit);

// Change PHI-node in loop header to use outer cond rather than preheader,
// and set IV to the LowerBound.
Instruction *Phi = &Header->front();
auto *PI = cast<PHINode>(Phi);
PI->setIncomingBlock(0, OuterCond);
PI->setIncomingValue(0, LowerBound);

// Then set the pre-header to jump to the OuterCond
Instruction *Term = PreHeader->getTerminator();
auto *Br = cast<BranchInst>(Term);
Br->setSuccessor(0, OuterCond);

// Modify the inner condition:
// * Use the UpperBound returned from the DynamicNext call.
// * jump to the loop outer loop when done with one of the inner loops.
Builder.SetInsertPoint(Cond, Cond->getFirstInsertionPt());
UpperBound = Builder.CreateLoad(IVTy, PUpperBound, "ub");
Instruction *Comp = &*Builder.GetInsertPoint();
auto *CI = cast<CmpInst>(Comp);
CI->setOperand(1, UpperBound);
// Redirect the inner exit to branch to outer condition.
Instruction *Branch = &Cond->back();
auto *BI = cast<BranchInst>(Branch);
assert(BI->getSuccessor(1) == Exit)(static_cast <bool> (BI->getSuccessor(1) == Exit) ? void
 (0) : __assert_fail ("BI->getSuccessor(1) == Exit", "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp"
, 2553, __extension__ __PRETTY_FUNCTION__));
BI->setSuccessor(1, OuterCond);

// Call the "fini" function if "ordered" is present in wsloop directive.
if (Ordered) {
  Builder.SetInsertPoint(&Latch->back());
  FunctionCallee DynamicFini = getKmpcForDynamicFiniForType(IVTy, M, *this);
  Builder.CreateCall(DynamicFini, {SrcLoc, ThreadNum});
}

// Add the barrier if requested.
if (NeedsBarrier) {
  Builder.SetInsertPoint(&Exit->back());
  createBarrier(LocationDescription(Builder.saveIP(), DL),
                omp::Directive::OMPD_for, /* ForceSimpleCall */ false,
                /* CheckCancelFlag */ false);
}

CLI->invalidate();
return AfterIP;
2573}

2575/// Redirect all edges that branch to \p OldTarget to \p NewTarget. That is,
2576/// after this \p OldTarget will be orphaned.
2577static void redirectAllPredecessorsTo(BasicBlock *OldTarget,
                                    BasicBlock *NewTarget, DebugLoc DL) {
for (BasicBlock *Pred : make_early_inc_range(predecessors(OldTarget)))
  redirectTo(Pred, NewTarget, DL);
2581}

2583/// Determine which blocks in \p BBs are reachable from outside and remove the
2584/// ones that are not reachable from the function.
2585static void removeUnusedBlocksFromParent(ArrayRef<BasicBlock *> BBs) {
SmallPtrSet<BasicBlock *, 6> BBsToErase{BBs.begin(), BBs.end()};
auto HasRemainingUses = [&BBsToErase](BasicBlock *BB) {
  for (Use &U : BB->uses()) {
    auto *UseInst = dyn_cast<Instruction>(U.getUser());
    if (!UseInst)
      continue;
    if (BBsToErase.count(UseInst->getParent()))
      continue;
    return true;
  }
  return false;
};

while (true) {
  bool Changed = false;
  for (BasicBlock *BB : make_early_inc_range(BBsToErase)) {
    if (HasRemainingUses(BB)) {
      BBsToErase.erase(BB);
      Changed = true;
    }
  }
  if (!Changed)
    break;
}

SmallVector<BasicBlock *, 7> BBVec(BBsToErase.begin(), BBsToErase.end());
DeleteDeadBlocks(BBVec);
2613}

2615CanonicalLoopInfo *
2616OpenMPIRBuilder::collapseLoops(DebugLoc DL, ArrayRef<CanonicalLoopInfo *> Loops,
                             InsertPointTy ComputeIP) {
assert(Loops.size() >= 1 && "At least one loop required")(static_cast <bool> (Loops.size() >= 1 && "At least one loop required"
) ? void (0) : __assert_fail ("Loops.size() >= 1 && \"At least one loop required\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2618, __extension__
 __PRETTY_FUNCTION__));
1
Assuming the condition is true→
2
←
'?' condition is true→
size_t NumLoops = Loops.size();

// Nothing to do if there is already just one loop.
if (NumLoops == 1)
3
←
Assuming 'NumLoops' is not equal to 1→
4
←
Taking false branch→
  return Loops.front();

CanonicalLoopInfo *Outermost = Loops.front();
CanonicalLoopInfo *Innermost = Loops.back();
BasicBlock *OrigPreheader = Outermost->getPreheader();
BasicBlock *OrigAfter = Outermost->getAfter();
Function *F = OrigPreheader->getParent();

// Loop control blocks that may become orphaned later.
SmallVector<BasicBlock *, 12> OldControlBBs;
OldControlBBs.reserve(6 * Loops.size());
for (CanonicalLoopInfo *Loop : Loops)
5
←
Assuming '__begin1' is equal to '__end1'→
  Loop->collectControlBlocks(OldControlBBs);

// Setup the IRBuilder for inserting the trip count computation.
Builder.SetCurrentDebugLocation(DL);
if (ComputeIP.isSet())
6
←
Taking true branch→
  Builder.restoreIP(ComputeIP);
else
  Builder.restoreIP(Outermost->getPreheaderIP());

// Derive the collapsed' loop trip count.
// TODO: Find common/largest indvar type.
Value *CollapsedTripCount = nullptr;
7
←
'CollapsedTripCount' initialized to a null pointer value→
for (CanonicalLoopInfo *L : Loops) {
8
←
Assuming '__begin1' is equal to '__end1'→
  assert(L->isValid() &&(static_cast <bool> (L->isValid() && "All loops to collapse must be valid canonical loops"
) ? void (0) : __assert_fail ("L->isValid() && \"All loops to collapse must be valid canonical loops\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2649, __extension__
 __PRETTY_FUNCTION__))
         "All loops to collapse must be valid canonical loops")(static_cast <bool> (L->isValid() && "All loops to collapse must be valid canonical loops"
) ? void (0) : __assert_fail ("L->isValid() && \"All loops to collapse must be valid canonical loops\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2649, __extension__
 __PRETTY_FUNCTION__));
  Value *OrigTripCount = L->getTripCount();
  if (!CollapsedTripCount) {
    CollapsedTripCount = OrigTripCount;
    continue;
  }

  // TODO: Enable UndefinedSanitizer to diagnose an overflow here.
  CollapsedTripCount = Builder.CreateMul(CollapsedTripCount, OrigTripCount,
                                         {}, /*HasNUW=*/true);
}

// Create the collapsed loop control flow.
CanonicalLoopInfo *Result =
    createLoopSkeleton(DL, CollapsedTripCount, F,
9
←
Passing null pointer value via 2nd parameter 'TripCount'→
10
←
Calling 'OpenMPIRBuilder::createLoopSkeleton'→
                       OrigPreheader->getNextNode(), OrigAfter, "collapsed");

// Build the collapsed loop body code.
// Start with deriving the input loop induction variables from the collapsed
// one, using a divmod scheme. To preserve the original loops' order, the
// innermost loop use the least significant bits.
Builder.restoreIP(Result->getBodyIP());

Value *Leftover = Result->getIndVar();
SmallVector<Value *> NewIndVars;
NewIndVars.resize(NumLoops);
for (int i = NumLoops - 1; i >= 1; --i) {
  Value *OrigTripCount = Loops[i]->getTripCount();

  Value *NewIndVar = Builder.CreateURem(Leftover, OrigTripCount);
  NewIndVars[i] = NewIndVar;

  Leftover = Builder.CreateUDiv(Leftover, OrigTripCount);
}
// Outermost loop gets all the remaining bits.
NewIndVars[0] = Leftover;

// Construct the loop body control flow.
// We progressively construct the branch structure following in direction of
// the control flow, from the leading in-between code, the loop nest body, the
// trailing in-between code, and rejoining the collapsed loop's latch.
// ContinueBlock and ContinuePred keep track of the source(s) of next edge. If
// the ContinueBlock is set, continue with that block. If ContinuePred, use
// its predecessors as sources.
BasicBlock *ContinueBlock = Result->getBody();
BasicBlock *ContinuePred = nullptr;
auto ContinueWith = [&ContinueBlock, &ContinuePred, DL](BasicBlock *Dest,
                                                        BasicBlock *NextSrc) {
  if (ContinueBlock)
    redirectTo(ContinueBlock, Dest, DL);
  else
    redirectAllPredecessorsTo(ContinuePred, Dest, DL);

  ContinueBlock = nullptr;
  ContinuePred = NextSrc;
};

// The code before the nested loop of each level.
// Because we are sinking it into the nest, it will be executed more often
// that the original loop. More sophisticated schemes could keep track of what
// the in-between code is and instantiate it only once per thread.
for (size_t i = 0; i < NumLoops - 1; ++i)
  ContinueWith(Loops[i]->getBody(), Loops[i + 1]->getHeader());

// Connect the loop nest body.
ContinueWith(Innermost->getBody(), Innermost->getLatch());

// The code after the nested loop at each level.
for (size_t i = NumLoops - 1; i > 0; --i)
  ContinueWith(Loops[i]->getAfter(), Loops[i - 1]->getLatch());

// Connect the finished loop to the collapsed loop latch.
ContinueWith(Result->getLatch(), nullptr);

// Replace the input loops with the new collapsed loop.
redirectTo(Outermost->getPreheader(), Result->getPreheader(), DL);
redirectTo(Result->getAfter(), Outermost->getAfter(), DL);

// Replace the input loop indvars with the derived ones.
for (size_t i = 0; i < NumLoops; ++i)
  Loops[i]->getIndVar()->replaceAllUsesWith(NewIndVars[i]);

// Remove unused parts of the input loops.
removeUnusedBlocksFromParent(OldControlBBs);

for (CanonicalLoopInfo *L : Loops)
  L->invalidate();

2737#ifndef NDEBUG
Result->assertOK();
2739#endif
return Result;
2741}

2743std::vector<CanonicalLoopInfo *>
2744OpenMPIRBuilder::tileLoops(DebugLoc DL, ArrayRef<CanonicalLoopInfo *> Loops,
                         ArrayRef<Value *> TileSizes) {
assert(TileSizes.size() == Loops.size() &&(static_cast <bool> (TileSizes.size() == Loops.size() &&
 "Must pass as many tile sizes as there are loops") ? void (0
) : __assert_fail ("TileSizes.size() == Loops.size() && \"Must pass as many tile sizes as there are loops\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2747, __extension__
 __PRETTY_FUNCTION__))
       "Must pass as many tile sizes as there are loops")(static_cast <bool> (TileSizes.size() == Loops.size() &&
 "Must pass as many tile sizes as there are loops") ? void (0
) : __assert_fail ("TileSizes.size() == Loops.size() && \"Must pass as many tile sizes as there are loops\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2747, __extension__
 __PRETTY_FUNCTION__));
int NumLoops = Loops.size();
assert(NumLoops >= 1 && "At least one loop to tile required")(static_cast <bool> (NumLoops >= 1 && "At least one loop to tile required"
) ? void (0) : __assert_fail ("NumLoops >= 1 && \"At least one loop to tile required\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2749, __extension__
 __PRETTY_FUNCTION__));

CanonicalLoopInfo *OutermostLoop = Loops.front();
CanonicalLoopInfo *InnermostLoop = Loops.back();
Function *F = OutermostLoop->getBody()->getParent();
BasicBlock *InnerEnter = InnermostLoop->getBody();
BasicBlock *InnerLatch = InnermostLoop->getLatch();

// Loop control blocks that may become orphaned later.
SmallVector<BasicBlock *, 12> OldControlBBs;
OldControlBBs.reserve(6 * Loops.size());
for (CanonicalLoopInfo *Loop : Loops)
  Loop->collectControlBlocks(OldControlBBs);

// Collect original trip counts and induction variable to be accessible by
// index. Also, the structure of the original loops is not preserved during
// the construction of the tiled loops, so do it before we scavenge the BBs of
// any original CanonicalLoopInfo.
SmallVector<Value *, 4> OrigTripCounts, OrigIndVars;
for (CanonicalLoopInfo *L : Loops) {
  assert(L->isValid() && "All input loops must be valid canonical loops")(static_cast <bool> (L->isValid() && "All input loops must be valid canonical loops"
) ? void (0) : __assert_fail ("L->isValid() && \"All input loops must be valid canonical loops\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2769, __extension__
 __PRETTY_FUNCTION__));
  OrigTripCounts.push_back(L->getTripCount());
  OrigIndVars.push_back(L->getIndVar());
}

// Collect the code between loop headers. These may contain SSA definitions
// that are used in the loop nest body. To be usable with in the innermost
// body, these BasicBlocks will be sunk into the loop nest body. That is,
// these instructions may be executed more often than before the tiling.
// TODO: It would be sufficient to only sink them into body of the
// corresponding tile loop.
SmallVector<std::pair<BasicBlock *, BasicBlock *>, 4> InbetweenCode;
for (int i = 0; i < NumLoops - 1; ++i) {
  CanonicalLoopInfo *Surrounding = Loops[i];
  CanonicalLoopInfo *Nested = Loops[i + 1];

  BasicBlock *EnterBB = Surrounding->getBody();
  BasicBlock *ExitBB = Nested->getHeader();
  InbetweenCode.emplace_back(EnterBB, ExitBB);
}

// Compute the trip counts of the floor loops.
Builder.SetCurrentDebugLocation(DL);
Builder.restoreIP(OutermostLoop->getPreheaderIP());
SmallVector<Value *, 4> FloorCount, FloorRems;
for (int i = 0; i < NumLoops; ++i) {
  Value *TileSize = TileSizes[i];
  Value *OrigTripCount = OrigTripCounts[i];
  Type *IVType = OrigTripCount->getType();

  Value *FloorTripCount = Builder.CreateUDiv(OrigTripCount, TileSize);
  Value *FloorTripRem = Builder.CreateURem(OrigTripCount, TileSize);

  // 0 if tripcount divides the tilesize, 1 otherwise.
  // 1 means we need an additional iteration for a partial tile.
  //
  // Unfortunately we cannot just use the roundup-formula
  //   (tripcount + tilesize - 1)/tilesize
  // because the summation might overflow. We do not want introduce undefined
  // behavior when the untiled loop nest did not.
  Value *FloorTripOverflow =
      Builder.CreateICmpNE(FloorTripRem, ConstantInt::get(IVType, 0));

  FloorTripOverflow = Builder.CreateZExt(FloorTripOverflow, IVType);
  FloorTripCount =
      Builder.CreateAdd(FloorTripCount, FloorTripOverflow,
                        "omp_floor" + Twine(i) + ".tripcount", true);

  // Remember some values for later use.
  FloorCount.push_back(FloorTripCount);
  FloorRems.push_back(FloorTripRem);
}

// Generate the new loop nest, from the outermost to the innermost.
std::vector<CanonicalLoopInfo *> Result;
Result.reserve(NumLoops * 2);

// The basic block of the surrounding loop that enters the nest generated
// loop.
BasicBlock *Enter = OutermostLoop->getPreheader();

// The basic block of the surrounding loop where the inner code should
// continue.
BasicBlock *Continue = OutermostLoop->getAfter();

// Where the next loop basic block should be inserted.
BasicBlock *OutroInsertBefore = InnermostLoop->getExit();

auto EmbeddNewLoop =
    [this, DL, F, InnerEnter, &Enter, &Continue, &OutroInsertBefore](
        Value *TripCount, const Twine &Name) -> CanonicalLoopInfo * {
  CanonicalLoopInfo *EmbeddedLoop = createLoopSkeleton(
      DL, TripCount, F, InnerEnter, OutroInsertBefore, Name);
  redirectTo(Enter, EmbeddedLoop->getPreheader(), DL);
  redirectTo(EmbeddedLoop->getAfter(), Continue, DL);

  // Setup the position where the next embedded loop connects to this loop.
  Enter = EmbeddedLoop->getBody();
  Continue = EmbeddedLoop->getLatch();
  OutroInsertBefore = EmbeddedLoop->getLatch();
  return EmbeddedLoop;
};

auto EmbeddNewLoops = [&Result, &EmbeddNewLoop](ArrayRef<Value *> TripCounts,
                                                const Twine &NameBase) {
  for (auto P : enumerate(TripCounts)) {
    CanonicalLoopInfo *EmbeddedLoop =
        EmbeddNewLoop(P.value(), NameBase + Twine(P.index()));
    Result.push_back(EmbeddedLoop);
  }
};

EmbeddNewLoops(FloorCount, "floor");

// Within the innermost floor loop, emit the code that computes the tile
// sizes.
Builder.SetInsertPoint(Enter->getTerminator());
SmallVector<Value *, 4> TileCounts;
for (int i = 0; i < NumLoops; ++i) {
  CanonicalLoopInfo *FloorLoop = Result[i];
  Value *TileSize = TileSizes[i];

  Value *FloorIsEpilogue =
      Builder.CreateICmpEQ(FloorLoop->getIndVar(), FloorCount[i]);
  Value *TileTripCount =
      Builder.CreateSelect(FloorIsEpilogue, FloorRems[i], TileSize);

  TileCounts.push_back(TileTripCount);
}

// Create the tile loops.
EmbeddNewLoops(TileCounts, "tile");

// Insert the inbetween code into the body.
BasicBlock *BodyEnter = Enter;
BasicBlock *BodyEntered = nullptr;
for (std::pair<BasicBlock *, BasicBlock *> P : InbetweenCode) {
  BasicBlock *EnterBB = P.first;
  BasicBlock *ExitBB = P.second;

  if (BodyEnter)
    redirectTo(BodyEnter, EnterBB, DL);
  else
    redirectAllPredecessorsTo(BodyEntered, EnterBB, DL);

  BodyEnter = nullptr;
  BodyEntered = ExitBB;
}

// Append the original loop nest body into the generated loop nest body.
if (BodyEnter)
  redirectTo(BodyEnter, InnerEnter, DL);
else
  redirectAllPredecessorsTo(BodyEntered, InnerEnter, DL);
redirectAllPredecessorsTo(InnerLatch, Continue, DL);

// Replace the original induction variable with an induction variable computed
// from the tile and floor induction variables.
Builder.restoreIP(Result.back()->getBodyIP());
for (int i = 0; i < NumLoops; ++i) {
  CanonicalLoopInfo *FloorLoop = Result[i];
  CanonicalLoopInfo *TileLoop = Result[NumLoops + i];
  Value *OrigIndVar = OrigIndVars[i];
  Value *Size = TileSizes[i];

  Value *Scale =
      Builder.CreateMul(Size, FloorLoop->getIndVar(), {}, /*HasNUW=*/true);
  Value *Shift =
      Builder.CreateAdd(Scale, TileLoop->getIndVar(), {}, /*HasNUW=*/true);
  OrigIndVar->replaceAllUsesWith(Shift);
}

// Remove unused parts of the original loops.
removeUnusedBlocksFromParent(OldControlBBs);

for (CanonicalLoopInfo *L : Loops)
  L->invalidate();

2927#ifndef NDEBUG
for (CanonicalLoopInfo *GenL : Result)
  GenL->assertOK();
2930#endif
return Result;
2932}

2934/// Attach metadata \p Properties to the basic block described by \p BB. If the
2935/// basic block already has metadata, the basic block properties are appended.
2936static void addBasicBlockMetadata(BasicBlock *BB,
                                ArrayRef<Metadata *> Properties) {
// Nothing to do if no property to attach.
if (Properties.empty())
  return;

LLVMContext &Ctx = BB->getContext();
SmallVector<Metadata *> NewProperties;
NewProperties.push_back(nullptr);

// If the basic block already has metadata, prepend it to the new metadata.
MDNode *Existing = BB->getTerminator()->getMetadata(LLVMContext::MD_loop);
if (Existing)
  append_range(NewProperties, drop_begin(Existing->operands(), 1));

append_range(NewProperties, Properties);
MDNode *BasicBlockID = MDNode::getDistinct(Ctx, NewProperties);
BasicBlockID->replaceOperandWith(0, BasicBlockID);

BB->getTerminator()->setMetadata(LLVMContext::MD_loop, BasicBlockID);
2956}

2958/// Attach loop metadata \p Properties to the loop described by \p Loop. If the
2959/// loop already has metadata, the loop properties are appended.
2960static void addLoopMetadata(CanonicalLoopInfo *Loop,
                          ArrayRef<Metadata *> Properties) {
assert(Loop->isValid() && "Expecting a valid CanonicalLoopInfo")(static_cast <bool> (Loop->isValid() && "Expecting a valid CanonicalLoopInfo"
) ? void (0) : __assert_fail ("Loop->isValid() && \"Expecting a valid CanonicalLoopInfo\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2962, __extension__
 __PRETTY_FUNCTION__));

// Attach metadata to the loop's latch
BasicBlock *Latch = Loop->getLatch();
assert(Latch && "A valid CanonicalLoopInfo must have a unique latch")(static_cast <bool> (Latch && "A valid CanonicalLoopInfo must have a unique latch"
) ? void (0) : __assert_fail ("Latch && \"A valid CanonicalLoopInfo must have a unique latch\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 2966, __extension__
 __PRETTY_FUNCTION__));
addBasicBlockMetadata(Latch, Properties);
2968}

2970/// Attach llvm.access.group metadata to the memref instructions of \p Block
2971static void addSimdMetadata(BasicBlock *Block, MDNode *AccessGroup,
                          LoopInfo &LI) {
for (Instruction &I : *Block) {
  if (I.mayReadOrWriteMemory()) {
    // TODO: This instruction may already have access group from
    // other pragmas e.g. #pragma clang loop vectorize.  Append
    // so that the existing metadata is not overwritten.
    I.setMetadata(LLVMContext::MD_access_group, AccessGroup);
  }
}
2981}

2983void OpenMPIRBuilder::unrollLoopFull(DebugLoc, CanonicalLoopInfo *Loop) {
LLVMContext &Ctx = Builder.getContext();
addLoopMetadata(
    Loop, {MDNode::get(Ctx, MDString::get(Ctx, "llvm.loop.unroll.enable")),
           MDNode::get(Ctx, MDString::get(Ctx, "llvm.loop.unroll.full"))});
2988}

2990void OpenMPIRBuilder::unrollLoopHeuristic(DebugLoc, CanonicalLoopInfo *Loop) {
LLVMContext &Ctx = Builder.getContext();
addLoopMetadata(
    Loop, {
              MDNode::get(Ctx, MDString::get(Ctx, "llvm.loop.unroll.enable")),
          });
2996}

2998void OpenMPIRBuilder::createIfVersion(CanonicalLoopInfo *CanonicalLoop,
                                    Value *IfCond, ValueToValueMapTy &VMap,
                                    const Twine &NamePrefix) {
Function *F = CanonicalLoop->getFunction();

// Define where if branch should be inserted
Instruction *SplitBefore;
if (Instruction::classof(IfCond)) {
  SplitBefore = dyn_cast<Instruction>(IfCond);
} else {
  SplitBefore = CanonicalLoop->getPreheader()->getTerminator();
}

// TODO: We should not rely on pass manager. Currently we use pass manager
// only for getting llvm::Loop which corresponds to given CanonicalLoopInfo
// object. We should have a method  which returns all blocks between
// CanonicalLoopInfo::getHeader() and CanonicalLoopInfo::getAfter()
FunctionAnalysisManager FAM;
FAM.registerPass([]() { return DominatorTreeAnalysis(); });
FAM.registerPass([]() { return LoopAnalysis(); });
FAM.registerPass([]() { return PassInstrumentationAnalysis(); });

// Get the loop which needs to be cloned
LoopAnalysis LIA;
LoopInfo &&LI = LIA.run(*F, FAM);
Loop *L = LI.getLoopFor(CanonicalLoop->getHeader());

// Create additional blocks for the if statement
BasicBlock *Head = SplitBefore->getParent();
Instruction *HeadOldTerm = Head->getTerminator();
llvm::LLVMContext &C = Head->getContext();
llvm::BasicBlock *ThenBlock = llvm::BasicBlock::Create(
    C, NamePrefix + ".if.then", Head->getParent(), Head->getNextNode());
llvm::BasicBlock *ElseBlock = llvm::BasicBlock::Create(
    C, NamePrefix + ".if.else", Head->getParent(), CanonicalLoop->getExit());

// Create if condition branch.
Builder.SetInsertPoint(HeadOldTerm);
Instruction *BrInstr =
    Builder.CreateCondBr(IfCond, ThenBlock, /*ifFalse*/ ElseBlock);
InsertPointTy IP{BrInstr->getParent(), ++BrInstr->getIterator()};
// Then block contains branch to omp loop which needs to be vectorized
spliceBB(IP, ThenBlock, false);
ThenBlock->replaceSuccessorsPhiUsesWith(Head, ThenBlock);

Builder.SetInsertPoint(ElseBlock);

// Clone loop for the else branch
SmallVector<BasicBlock *, 8> NewBlocks;

VMap[CanonicalLoop->getPreheader()] = ElseBlock;
for (BasicBlock *Block : L->getBlocks()) {
  BasicBlock *NewBB = CloneBasicBlock(Block, VMap, "", F);
  NewBB->moveBefore(CanonicalLoop->getExit());
  VMap[Block] = NewBB;
  NewBlocks.push_back(NewBB);
}
remapInstructionsInBlocks(NewBlocks, VMap);
Builder.CreateBr(NewBlocks.front());
3057}

3059unsigned
3060OpenMPIRBuilder::getOpenMPDefaultSimdAlign(const Triple &TargetTriple,
                                         const StringMap<bool> &Features) {
if (TargetTriple.isX86()) {
  if (Features.lookup("avx512f"))
    return 512;
  else if (Features.lookup("avx"))
    return 256;
  return 128;
}
if (TargetTriple.isPPC())
  return 128;
if (TargetTriple.isWasm())
  return 128;
return 0;
3074}

3076void OpenMPIRBuilder::applySimd(CanonicalLoopInfo *CanonicalLoop,
                              MapVector<Value *, Value *> AlignedVars,
                              Value *IfCond, OrderKind Order,
                              ConstantInt *Simdlen, ConstantInt *Safelen) {
LLVMContext &Ctx = Builder.getContext();

Function *F = CanonicalLoop->getFunction();

// TODO: We should not rely on pass manager. Currently we use pass manager
// only for getting llvm::Loop which corresponds to given CanonicalLoopInfo
// object. We should have a method  which returns all blocks between
// CanonicalLoopInfo::getHeader() and CanonicalLoopInfo::getAfter()
FunctionAnalysisManager FAM;
FAM.registerPass([]() { return DominatorTreeAnalysis(); });
FAM.registerPass([]() { return LoopAnalysis(); });
FAM.registerPass([]() { return PassInstrumentationAnalysis(); });

LoopAnalysis LIA;
LoopInfo &&LI = LIA.run(*F, FAM);

Loop *L = LI.getLoopFor(CanonicalLoop->getHeader());
if (AlignedVars.size()) {
  InsertPointTy IP = Builder.saveIP();
  Builder.SetInsertPoint(CanonicalLoop->getPreheader()->getTerminator());
  for (auto &AlignedItem : AlignedVars) {
    Value *AlignedPtr = AlignedItem.first;
    Value *Alignment = AlignedItem.second;
    Builder.CreateAlignmentAssumption(F->getParent()->getDataLayout(),
                                      AlignedPtr, Alignment);
  }
  Builder.restoreIP(IP);
}

if (IfCond) {
  ValueToValueMapTy VMap;
  createIfVersion(CanonicalLoop, IfCond, VMap, "simd");
  // Add metadata to the cloned loop which disables vectorization
  Value *MappedLatch = VMap.lookup(CanonicalLoop->getLatch());
  assert(MappedLatch &&(static_cast <bool> (MappedLatch && "Cannot find value which corresponds to original loop latch"
) ? void (0) : __assert_fail ("MappedLatch && \"Cannot find value which corresponds to original loop latch\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3115, __extension__
 __PRETTY_FUNCTION__))
         "Cannot find value which corresponds to original loop latch")(static_cast <bool> (MappedLatch && "Cannot find value which corresponds to original loop latch"
) ? void (0) : __assert_fail ("MappedLatch && \"Cannot find value which corresponds to original loop latch\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3115, __extension__
 __PRETTY_FUNCTION__));
  assert(isa<BasicBlock>(MappedLatch) &&(static_cast <bool> (isa<BasicBlock>(MappedLatch)
 && "Cannot cast mapped latch block value to BasicBlock"
) ? void (0) : __assert_fail ("isa<BasicBlock>(MappedLatch) && \"Cannot cast mapped latch block value to BasicBlock\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3117, __extension__
 __PRETTY_FUNCTION__))
         "Cannot cast mapped latch block value to BasicBlock")(static_cast <bool> (isa<BasicBlock>(MappedLatch)
 && "Cannot cast mapped latch block value to BasicBlock"
) ? void (0) : __assert_fail ("isa<BasicBlock>(MappedLatch) && \"Cannot cast mapped latch block value to BasicBlock\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3117, __extension__
 __PRETTY_FUNCTION__));
  BasicBlock *NewLatchBlock = dyn_cast<BasicBlock>(MappedLatch);
  ConstantAsMetadata *BoolConst =
      ConstantAsMetadata::get(ConstantInt::getFalse(Type::getInt1Ty(Ctx)));
  addBasicBlockMetadata(
      NewLatchBlock,
      {MDNode::get(Ctx, {MDString::get(Ctx, "llvm.loop.vectorize.enable"),
                         BoolConst})});
}

SmallSet<BasicBlock *, 8> Reachable;

// Get the basic blocks from the loop in which memref instructions
// can be found.
// TODO: Generalize getting all blocks inside a CanonicalizeLoopInfo,
// preferably without running any passes.
for (BasicBlock *Block : L->getBlocks()) {
  if (Block == CanonicalLoop->getCond() ||
      Block == CanonicalLoop->getHeader())
    continue;
  Reachable.insert(Block);
}

SmallVector<Metadata *> LoopMDList;

// In presence of finite 'safelen', it may be unsafe to mark all
// the memory instructions parallel, because loop-carried
// dependences of 'safelen' iterations are possible.
// If clause order(concurrent) is specified then the memory instructions
// are marked parallel even if 'safelen' is finite.
if ((Safelen == nullptr) || (Order == OrderKind::OMP_ORDER_concurrent)) {
  // Add access group metadata to memory-access instructions.
  MDNode *AccessGroup = MDNode::getDistinct(Ctx, {});
  for (BasicBlock *BB : Reachable)
    addSimdMetadata(BB, AccessGroup, LI);
  // TODO:  If the loop has existing parallel access metadata, have
  // to combine two lists.
  LoopMDList.push_back(MDNode::get(
      Ctx, {MDString::get(Ctx, "llvm.loop.parallel_accesses"), AccessGroup}));
}

// Use the above access group metadata to create loop level
// metadata, which should be distinct for each loop.
ConstantAsMetadata *BoolConst =
    ConstantAsMetadata::get(ConstantInt::getTrue(Type::getInt1Ty(Ctx)));
LoopMDList.push_back(MDNode::get(
    Ctx, {MDString::get(Ctx, "llvm.loop.vectorize.enable"), BoolConst}));

if (Simdlen || Safelen) {
  // If both simdlen and safelen clauses are specified, the value of the
  // simdlen parameter must be less than or equal to the value of the safelen
  // parameter. Therefore, use safelen only in the absence of simdlen.
  ConstantInt *VectorizeWidth = Simdlen == nullptr ? Safelen : Simdlen;
  LoopMDList.push_back(
      MDNode::get(Ctx, {MDString::get(Ctx, "llvm.loop.vectorize.width"),
                        ConstantAsMetadata::get(VectorizeWidth)}));
}

addLoopMetadata(CanonicalLoop, LoopMDList);
3176}

3178/// Create the TargetMachine object to query the backend for optimization
3179/// preferences.
3180///
3181/// Ideally, this would be passed from the front-end to the OpenMPBuilder, but
3182/// e.g. Clang does not pass it to its CodeGen layer and creates it only when
3183/// needed for the LLVM pass pipline. We use some default options to avoid
3184/// having to pass too many settings from the frontend that probably do not
3185/// matter.
3186///
3187/// Currently, TargetMachine is only used sometimes by the unrollLoopPartial
3188/// method. If we are going to use TargetMachine for more purposes, especially
3189/// those that are sensitive to TargetOptions, RelocModel and CodeModel, it
3190/// might become be worth requiring front-ends to pass on their TargetMachine,
3191/// or at least cache it between methods. Note that while fontends such as Clang
3192/// have just a single main TargetMachine per translation unit, "target-cpu" and
3193/// "target-features" that determine the TargetMachine are per-function and can
3194/// be overrided using __attribute__((target("OPTIONS"))).
3195static std::unique_ptr<TargetMachine>
3196createTargetMachine(Function *F, CodeGenOpt::Level OptLevel) {
Module *M = F->getParent();

StringRef CPU = F->getFnAttribute("target-cpu").getValueAsString();
StringRef Features = F->getFnAttribute("target-features").getValueAsString();
const std::string &Triple = M->getTargetTriple();

std::string Error;
const llvm::Target *TheTarget = TargetRegistry::lookupTarget(Triple, Error);
if (!TheTarget)
  return {};

llvm::TargetOptions Options;
return std::unique_ptr<TargetMachine>(TheTarget->createTargetMachine(
    Triple, CPU, Features, Options, /*RelocModel=*/std::nullopt,
    /*CodeModel=*/std::nullopt, OptLevel));
3212}

3214/// Heuristically determine the best-performant unroll factor for \p CLI. This
3215/// depends on the target processor. We are re-using the same heuristics as the
3216/// LoopUnrollPass.
3217static int32_t computeHeuristicUnrollFactor(CanonicalLoopInfo *CLI) {
Function *F = CLI->getFunction();

// Assume the user requests the most aggressive unrolling, even if the rest of
// the code is optimized using a lower setting.
CodeGenOpt::Level OptLevel = CodeGenOpt::Aggressive;
std::unique_ptr<TargetMachine> TM = createTargetMachine(F, OptLevel);

FunctionAnalysisManager FAM;
FAM.registerPass([]() { return TargetLibraryAnalysis(); });
FAM.registerPass([]() { return AssumptionAnalysis(); });
FAM.registerPass([]() { return DominatorTreeAnalysis(); });
FAM.registerPass([]() { return LoopAnalysis(); });
FAM.registerPass([]() { return ScalarEvolutionAnalysis(); });
FAM.registerPass([]() { return PassInstrumentationAnalysis(); });
TargetIRAnalysis TIRA;
if (TM)
  TIRA = TargetIRAnalysis(
      [&](const Function &F) { return TM->getTargetTransformInfo(F); });
FAM.registerPass([&]() { return TIRA; });

TargetIRAnalysis::Result &&TTI = TIRA.run(*F, FAM);
ScalarEvolutionAnalysis SEA;
ScalarEvolution &&SE = SEA.run(*F, FAM);
DominatorTreeAnalysis DTA;
DominatorTree &&DT = DTA.run(*F, FAM);
LoopAnalysis LIA;
LoopInfo &&LI = LIA.run(*F, FAM);
AssumptionAnalysis ACT;
AssumptionCache &&AC = ACT.run(*F, FAM);
OptimizationRemarkEmitter ORE{F};

Loop *L = LI.getLoopFor(CLI->getHeader());
assert(L && "Expecting CanonicalLoopInfo to be recognized as a loop")(static_cast <bool> (L && "Expecting CanonicalLoopInfo to be recognized as a loop"
) ? void (0) : __assert_fail ("L && \"Expecting CanonicalLoopInfo to be recognized as a loop\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3250, __extension__
 __PRETTY_FUNCTION__));

TargetTransformInfo::UnrollingPreferences UP =
    gatherUnrollingPreferences(L, SE, TTI,
                               /*BlockFrequencyInfo=*/nullptr,
                               /*ProfileSummaryInfo=*/nullptr, ORE, OptLevel,
                               /*UserThreshold=*/std::nullopt,
                               /*UserCount=*/std::nullopt,
                               /*UserAllowPartial=*/true,
                               /*UserAllowRuntime=*/true,
                               /*UserUpperBound=*/std::nullopt,
                               /*UserFullUnrollMaxCount=*/std::nullopt);

UP.Force = true;

// Account for additional optimizations taking place before the LoopUnrollPass
// would unroll the loop.
UP.Threshold *= UnrollThresholdFactor;
UP.PartialThreshold *= UnrollThresholdFactor;

// Use normal unroll factors even if the rest of the code is optimized for
// size.
UP.OptSizeThreshold = UP.Threshold;
UP.PartialOptSizeThreshold = UP.PartialThreshold;

LLVM_DEBUG(dbgs() << "Unroll heuristic thresholds:\n"do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Unroll heuristic thresholds:\n"
 << "  Threshold=" << UP.Threshold << "\n" <<
 "  PartialThreshold=" << UP.PartialThreshold << "\n"
 << "  OptSizeThreshold=" << UP.OptSizeThreshold <<
 "\n" << "  PartialOptSizeThreshold=" << UP.PartialOptSizeThreshold
 << "\n"; } } while (false)
                  << "  Threshold=" << UP.Threshold << "\n"do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Unroll heuristic thresholds:\n"
 << "  Threshold=" << UP.Threshold << "\n" <<
 "  PartialThreshold=" << UP.PartialThreshold << "\n"
 << "  OptSizeThreshold=" << UP.OptSizeThreshold <<
 "\n" << "  PartialOptSizeThreshold=" << UP.PartialOptSizeThreshold
 << "\n"; } } while (false)
                  << "  PartialThreshold=" << UP.PartialThreshold << "\n"do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Unroll heuristic thresholds:\n"
 << "  Threshold=" << UP.Threshold << "\n" <<
 "  PartialThreshold=" << UP.PartialThreshold << "\n"
 << "  OptSizeThreshold=" << UP.OptSizeThreshold <<
 "\n" << "  PartialOptSizeThreshold=" << UP.PartialOptSizeThreshold
 << "\n"; } } while (false)
                  << "  OptSizeThreshold=" << UP.OptSizeThreshold << "\n"do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Unroll heuristic thresholds:\n"
 << "  Threshold=" << UP.Threshold << "\n" <<
 "  PartialThreshold=" << UP.PartialThreshold << "\n"
 << "  OptSizeThreshold=" << UP.OptSizeThreshold <<
 "\n" << "  PartialOptSizeThreshold=" << UP.PartialOptSizeThreshold
 << "\n"; } } while (false)
                  << "  PartialOptSizeThreshold="do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Unroll heuristic thresholds:\n"
 << "  Threshold=" << UP.Threshold << "\n" <<
 "  PartialThreshold=" << UP.PartialThreshold << "\n"
 << "  OptSizeThreshold=" << UP.OptSizeThreshold <<
 "\n" << "  PartialOptSizeThreshold=" << UP.PartialOptSizeThreshold
 << "\n"; } } while (false)
                  << UP.PartialOptSizeThreshold << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Unroll heuristic thresholds:\n"
 << "  Threshold=" << UP.Threshold << "\n" <<
 "  PartialThreshold=" << UP.PartialThreshold << "\n"
 << "  OptSizeThreshold=" << UP.OptSizeThreshold <<
 "\n" << "  PartialOptSizeThreshold=" << UP.PartialOptSizeThreshold
 << "\n"; } } while (false);

// Disable peeling.
TargetTransformInfo::PeelingPreferences PP =
    gatherPeelingPreferences(L, SE, TTI,
                             /*UserAllowPeeling=*/false,
                             /*UserAllowProfileBasedPeeling=*/false,
                             /*UnrollingSpecficValues=*/false);

SmallPtrSet<const Value *, 32> EphValues;
CodeMetrics::collectEphemeralValues(L, &AC, EphValues);

// Assume that reads and writes to stack variables can be eliminated by
// Mem2Reg, SROA or LICM. That is, don't count them towards the loop body's
// size.
for (BasicBlock *BB : L->blocks()) {
  for (Instruction &I : *BB) {
    Value *Ptr;
    if (auto *Load = dyn_cast<LoadInst>(&I)) {
      Ptr = Load->getPointerOperand();
    } else if (auto *Store = dyn_cast<StoreInst>(&I)) {
      Ptr = Store->getPointerOperand();
    } else
      continue;

    Ptr = Ptr->stripPointerCasts();

    if (auto *Alloca = dyn_cast<AllocaInst>(Ptr)) {
      if (Alloca->getParent() == &F->getEntryBlock())
        EphValues.insert(&I);
    }
  }
}

unsigned NumInlineCandidates;
bool NotDuplicatable;
bool Convergent;
InstructionCost LoopSizeIC =
    ApproximateLoopSize(L, NumInlineCandidates, NotDuplicatable, Convergent,
                        TTI, EphValues, UP.BEInsns);
LLVM_DEBUG(dbgs() << "Estimated loop size is " << LoopSizeIC << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Estimated loop size is "
 << LoopSizeIC << "\n"; } } while (false);

// Loop is not unrollable if the loop contains certain instructions.
if (NotDuplicatable || Convergent || !LoopSizeIC.isValid()) {
  LLVM_DEBUG(dbgs() << "Loop not considered unrollable\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Loop not considered unrollable\n"
; } } while (false);
  return 1;
}
unsigned LoopSize = *LoopSizeIC.getValue();

// TODO: Determine trip count of \p CLI if constant, computeUnrollCount might
// be able to use it.
int TripCount = 0;
int MaxTripCount = 0;
bool MaxOrZero = false;
unsigned TripMultiple = 0;

bool UseUpperBound = false;
computeUnrollCount(L, TTI, DT, &LI, &AC, SE, EphValues, &ORE, TripCount,
                   MaxTripCount, MaxOrZero, TripMultiple, LoopSize, UP, PP,
                   UseUpperBound);
unsigned Factor = UP.Count;
LLVM_DEBUG(dbgs() << "Suggesting unroll factor of " << Factor << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { dbgs() << "Suggesting unroll factor of "
 << Factor << "\n"; } } while (false);

// This function returns 1 to signal to not unroll a loop.
if (Factor == 0)
  return 1;
return Factor;
3347}

3349void OpenMPIRBuilder::unrollLoopPartial(DebugLoc DL, CanonicalLoopInfo *Loop,
                                      int32_t Factor,
                                      CanonicalLoopInfo **UnrolledCLI) {
assert(Factor >= 0 && "Unroll factor must not be negative")(static_cast <bool> (Factor >= 0 && "Unroll factor must not be negative"
) ? void (0) : __assert_fail ("Factor >= 0 && \"Unroll factor must not be negative\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3352, __extension__
 __PRETTY_FUNCTION__));

Function *F = Loop->getFunction();
LLVMContext &Ctx = F->getContext();

// If the unrolled loop is not used for another loop-associated directive, it
// is sufficient to add metadata for the LoopUnrollPass.
if (!UnrolledCLI) {
  SmallVector<Metadata *, 2> LoopMetadata;
  LoopMetadata.push_back(
      MDNode::get(Ctx, MDString::get(Ctx, "llvm.loop.unroll.enable")));

  if (Factor >= 1) {
    ConstantAsMetadata *FactorConst = ConstantAsMetadata::get(
        ConstantInt::get(Type::getInt32Ty(Ctx), APInt(32, Factor)));
    LoopMetadata.push_back(MDNode::get(
        Ctx, {MDString::get(Ctx, "llvm.loop.unroll.count"), FactorConst}));
  }

  addLoopMetadata(Loop, LoopMetadata);
  return;
}

// Heuristically determine the unroll factor.
if (Factor == 0)
  Factor = computeHeuristicUnrollFactor(Loop);

// No change required with unroll factor 1.
if (Factor == 1) {
  *UnrolledCLI = Loop;
  return;
}

assert(Factor >= 2 &&(static_cast <bool> (Factor >= 2 && "unrolling only makes sense with a factor of 2 or larger"
) ? void (0) : __assert_fail ("Factor >= 2 && \"unrolling only makes sense with a factor of 2 or larger\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3386, __extension__
 __PRETTY_FUNCTION__))
       "unrolling only makes sense with a factor of 2 or larger")(static_cast <bool> (Factor >= 2 && "unrolling only makes sense with a factor of 2 or larger"
) ? void (0) : __assert_fail ("Factor >= 2 && \"unrolling only makes sense with a factor of 2 or larger\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3386, __extension__
 __PRETTY_FUNCTION__));

Type *IndVarTy = Loop->getIndVarType();

// Apply partial unrolling by tiling the loop by the unroll-factor, then fully
// unroll the inner loop.
Value *FactorVal =
    ConstantInt::get(IndVarTy, APInt(IndVarTy->getIntegerBitWidth(), Factor,
                                     /*isSigned=*/false));
std::vector<CanonicalLoopInfo *> LoopNest =
    tileLoops(DL, {Loop}, {FactorVal});
assert(LoopNest.size() == 2 && "Expect 2 loops after tiling")(static_cast <bool> (LoopNest.size() == 2 && "Expect 2 loops after tiling"
) ? void (0) : __assert_fail ("LoopNest.size() == 2 && \"Expect 2 loops after tiling\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3397, __extension__
 __PRETTY_FUNCTION__));
*UnrolledCLI = LoopNest[0];
CanonicalLoopInfo *InnerLoop = LoopNest[1];

// LoopUnrollPass can only fully unroll loops with constant trip count.
// Unroll by the unroll factor with a fallback epilog for the remainder
// iterations if necessary.
ConstantAsMetadata *FactorConst = ConstantAsMetadata::get(
    ConstantInt::get(Type::getInt32Ty(Ctx), APInt(32, Factor)));
addLoopMetadata(
    InnerLoop,
    {MDNode::get(Ctx, MDString::get(Ctx, "llvm.loop.unroll.enable")),
     MDNode::get(
         Ctx, {MDString::get(Ctx, "llvm.loop.unroll.count"), FactorConst})});

3412#ifndef NDEBUG
(*UnrolledCLI)->assertOK();
3414#endif
3415}

3417OpenMPIRBuilder::InsertPointTy
3418OpenMPIRBuilder::createCopyPrivate(const LocationDescription &Loc,
                                 llvm::Value *BufSize, llvm::Value *CpyBuf,
                                 llvm::Value *CpyFn, llvm::Value *DidIt) {
if (!updateToLocation(Loc))
  return Loc.IP;

uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
Value *ThreadId = getOrCreateThreadID(Ident);

llvm::Value *DidItLD = Builder.CreateLoad(Builder.getInt32Ty(), DidIt);

Value *Args[] = {Ident, ThreadId, BufSize, CpyBuf, CpyFn, DidItLD};

Function *Fn = getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_copyprivate);
Builder.CreateCall(Fn, Args);

return Builder.saveIP();
3437}

3439OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::createSingle(
  const LocationDescription &Loc, BodyGenCallbackTy BodyGenCB,
  FinalizeCallbackTy FiniCB, bool IsNowait, llvm::Value *DidIt) {

if (!updateToLocation(Loc))
  return Loc.IP;

// If needed (i.e. not null), initialize `DidIt` with 0
if (DidIt) {
  Builder.CreateStore(Builder.getInt32(0), DidIt);
}

Directive OMPD = Directive::OMPD_single;
uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
Value *ThreadId = getOrCreateThreadID(Ident);
Value *Args[] = {Ident, ThreadId};

Function *EntryRTLFn = getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_single);
Instruction *EntryCall = Builder.CreateCall(EntryRTLFn, Args);

Function *ExitRTLFn = getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_end_single);
Instruction *ExitCall = Builder.CreateCall(ExitRTLFn, Args);

// generates the following:
// if (__kmpc_single()) {
//		.... single region ...
// 		__kmpc_end_single
// }
// __kmpc_barrier

EmitOMPInlinedRegion(OMPD, EntryCall, ExitCall, BodyGenCB, FiniCB,
                     /*Conditional*/ true,
                     /*hasFinalize*/ true);
if (!IsNowait)
  createBarrier(LocationDescription(Builder.saveIP(), Loc.DL),
                omp::Directive::OMPD_unknown, /* ForceSimpleCall */ false,
                /* CheckCancelFlag */ false);
return Builder.saveIP();
3479}

3481OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::createCritical(
  const LocationDescription &Loc, BodyGenCallbackTy BodyGenCB,
  FinalizeCallbackTy FiniCB, StringRef CriticalName, Value *HintInst) {

if (!updateToLocation(Loc))
  return Loc.IP;

Directive OMPD = Directive::OMPD_critical;
uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
Value *ThreadId = getOrCreateThreadID(Ident);
Value *LockVar = getOMPCriticalRegionLock(CriticalName);
Value *Args[] = {Ident, ThreadId, LockVar};

SmallVector<llvm::Value *, 4> EnterArgs(std::begin(Args), std::end(Args));
Function *RTFn = nullptr;
if (HintInst) {
  // Add Hint to entry Args and create call
  EnterArgs.push_back(HintInst);
  RTFn = getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_critical_with_hint);
} else {
  RTFn = getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_critical);
}
Instruction *EntryCall = Builder.CreateCall(RTFn, EnterArgs);

Function *ExitRTLFn =
    getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_end_critical);
Instruction *ExitCall = Builder.CreateCall(ExitRTLFn, Args);

return EmitOMPInlinedRegion(OMPD, EntryCall, ExitCall, BodyGenCB, FiniCB,
                            /*Conditional*/ false, /*hasFinalize*/ true);
3513}

3515OpenMPIRBuilder::InsertPointTy
3516OpenMPIRBuilder::createOrderedDepend(const LocationDescription &Loc,
                                   InsertPointTy AllocaIP, unsigned NumLoops,
                                   ArrayRef<llvm::Value *> StoreValues,
                                   const Twine &Name, bool IsDependSource) {
assert((static_cast <bool> (llvm::all_of(StoreValues, [](Value
 *SV) { return SV->getType()->isIntegerTy(64); }) &&
 "OpenMP runtime requires depend vec with i64 type") ? void (
0) : __assert_fail ("llvm::all_of(StoreValues, [](Value *SV) { return SV->getType()->isIntegerTy(64); }) && \"OpenMP runtime requires depend vec with i64 type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3523, __extension__
 __PRETTY_FUNCTION__))
    llvm::all_of(StoreValues,(static_cast <bool> (llvm::all_of(StoreValues, [](Value
 *SV) { return SV->getType()->isIntegerTy(64); }) &&
 "OpenMP runtime requires depend vec with i64 type") ? void (
0) : __assert_fail ("llvm::all_of(StoreValues, [](Value *SV) { return SV->getType()->isIntegerTy(64); }) && \"OpenMP runtime requires depend vec with i64 type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3523, __extension__
 __PRETTY_FUNCTION__))
                 [](Value *SV) { return SV->getType()->isIntegerTy(64); }) &&(static_cast <bool> (llvm::all_of(StoreValues, [](Value
 *SV) { return SV->getType()->isIntegerTy(64); }) &&
 "OpenMP runtime requires depend vec with i64 type") ? void (
0) : __assert_fail ("llvm::all_of(StoreValues, [](Value *SV) { return SV->getType()->isIntegerTy(64); }) && \"OpenMP runtime requires depend vec with i64 type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3523, __extension__
 __PRETTY_FUNCTION__))
    "OpenMP runtime requires depend vec with i64 type")(static_cast <bool> (llvm::all_of(StoreValues, [](Value
 *SV) { return SV->getType()->isIntegerTy(64); }) &&
 "OpenMP runtime requires depend vec with i64 type") ? void (
0) : __assert_fail ("llvm::all_of(StoreValues, [](Value *SV) { return SV->getType()->isIntegerTy(64); }) && \"OpenMP runtime requires depend vec with i64 type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3523, __extension__
 __PRETTY_FUNCTION__));

if (!updateToLocation(Loc))
  return Loc.IP;

// Allocate space for vector and generate alloc instruction.
auto *ArrI64Ty = ArrayType::get(Int64, NumLoops);
Builder.restoreIP(AllocaIP);
AllocaInst *ArgsBase = Builder.CreateAlloca(ArrI64Ty, nullptr, Name);
ArgsBase->setAlignment(Align(8));
Builder.restoreIP(Loc.IP);

// Store the index value with offset in depend vector.
for (unsigned I = 0; I < NumLoops; ++I) {
  Value *DependAddrGEPIter = Builder.CreateInBoundsGEP(
      ArrI64Ty, ArgsBase, {Builder.getInt64(0), Builder.getInt64(I)});
  StoreInst *STInst = Builder.CreateStore(StoreValues[I], DependAddrGEPIter);
  STInst->setAlignment(Align(8));
}

Value *DependBaseAddrGEP = Builder.CreateInBoundsGEP(
    ArrI64Ty, ArgsBase, {Builder.getInt64(0), Builder.getInt64(0)});

uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
Value *ThreadId = getOrCreateThreadID(Ident);
Value *Args[] = {Ident, ThreadId, DependBaseAddrGEP};

Function *RTLFn = nullptr;
if (IsDependSource)
  RTLFn = getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_doacross_post);
else
  RTLFn = getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_doacross_wait);
Builder.CreateCall(RTLFn, Args);

return Builder.saveIP();
3560}

3562OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::createOrderedThreadsSimd(
  const LocationDescription &Loc, BodyGenCallbackTy BodyGenCB,
  FinalizeCallbackTy FiniCB, bool IsThreads) {
if (!updateToLocation(Loc))
  return Loc.IP;

Directive OMPD = Directive::OMPD_ordered;
Instruction *EntryCall = nullptr;
Instruction *ExitCall = nullptr;

if (IsThreads) {
  uint32_t SrcLocStrSize;
  Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
  Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
  Value *ThreadId = getOrCreateThreadID(Ident);
  Value *Args[] = {Ident, ThreadId};

  Function *EntryRTLFn = getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_ordered);
  EntryCall = Builder.CreateCall(EntryRTLFn, Args);

  Function *ExitRTLFn =
      getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_end_ordered);
  ExitCall = Builder.CreateCall(ExitRTLFn, Args);
}

return EmitOMPInlinedRegion(OMPD, EntryCall, ExitCall, BodyGenCB, FiniCB,
                            /*Conditional*/ false, /*hasFinalize*/ true);
3589}

3591OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::EmitOMPInlinedRegion(
  Directive OMPD, Instruction *EntryCall, Instruction *ExitCall,
  BodyGenCallbackTy BodyGenCB, FinalizeCallbackTy FiniCB, bool Conditional,
  bool HasFinalize, bool IsCancellable) {

if (HasFinalize)
  FinalizationStack.push_back({FiniCB, OMPD, IsCancellable});

// Create inlined region's entry and body blocks, in preparation
// for conditional creation
BasicBlock *EntryBB = Builder.GetInsertBlock();
Instruction *SplitPos = EntryBB->getTerminator();
if (!isa_and_nonnull<BranchInst>(SplitPos))
  SplitPos = new UnreachableInst(Builder.getContext(), EntryBB);
BasicBlock *ExitBB = EntryBB->splitBasicBlock(SplitPos, "omp_region.end");
BasicBlock *FiniBB =
    EntryBB->splitBasicBlock(EntryBB->getTerminator(), "omp_region.finalize");

Builder.SetInsertPoint(EntryBB->getTerminator());
emitCommonDirectiveEntry(OMPD, EntryCall, ExitBB, Conditional);

// generate body
BodyGenCB(/* AllocaIP */ InsertPointTy(),
          /* CodeGenIP */ Builder.saveIP());

// emit exit call and do any needed finalization.
auto FinIP = InsertPointTy(FiniBB, FiniBB->getFirstInsertionPt());
assert(FiniBB->getTerminator()->getNumSuccessors() == 1 &&(static_cast <bool> (FiniBB->getTerminator()->getNumSuccessors
() == 1 && FiniBB->getTerminator()->getSuccessor
(0) == ExitBB && "Unexpected control flow graph state!!"
) ? void (0) : __assert_fail ("FiniBB->getTerminator()->getNumSuccessors() == 1 && FiniBB->getTerminator()->getSuccessor(0) == ExitBB && \"Unexpected control flow graph state!!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3620, __extension__
 __PRETTY_FUNCTION__))
       FiniBB->getTerminator()->getSuccessor(0) == ExitBB &&(static_cast <bool> (FiniBB->getTerminator()->getNumSuccessors
() == 1 && FiniBB->getTerminator()->getSuccessor
(0) == ExitBB && "Unexpected control flow graph state!!"
) ? void (0) : __assert_fail ("FiniBB->getTerminator()->getNumSuccessors() == 1 && FiniBB->getTerminator()->getSuccessor(0) == ExitBB && \"Unexpected control flow graph state!!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3620, __extension__
 __PRETTY_FUNCTION__))
       "Unexpected control flow graph state!!")(static_cast <bool> (FiniBB->getTerminator()->getNumSuccessors
() == 1 && FiniBB->getTerminator()->getSuccessor
(0) == ExitBB && "Unexpected control flow graph state!!"
) ? void (0) : __assert_fail ("FiniBB->getTerminator()->getNumSuccessors() == 1 && FiniBB->getTerminator()->getSuccessor(0) == ExitBB && \"Unexpected control flow graph state!!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3620, __extension__
 __PRETTY_FUNCTION__));
emitCommonDirectiveExit(OMPD, FinIP, ExitCall, HasFinalize);
assert(FiniBB->getUniquePredecessor()->getUniqueSuccessor() == FiniBB &&(static_cast <bool> (FiniBB->getUniquePredecessor()->
getUniqueSuccessor() == FiniBB && "Unexpected Control Flow State!"
) ? void (0) : __assert_fail ("FiniBB->getUniquePredecessor()->getUniqueSuccessor() == FiniBB && \"Unexpected Control Flow State!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3623, __extension__
 __PRETTY_FUNCTION__))
       "Unexpected Control Flow State!")(static_cast <bool> (FiniBB->getUniquePredecessor()->
getUniqueSuccessor() == FiniBB && "Unexpected Control Flow State!"
) ? void (0) : __assert_fail ("FiniBB->getUniquePredecessor()->getUniqueSuccessor() == FiniBB && \"Unexpected Control Flow State!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3623, __extension__
 __PRETTY_FUNCTION__));
MergeBlockIntoPredecessor(FiniBB);

// If we are skipping the region of a non conditional, remove the exit
// block, and clear the builder's insertion point.
assert(SplitPos->getParent() == ExitBB &&(static_cast <bool> (SplitPos->getParent() == ExitBB
 && "Unexpected Insertion point location!") ? void (0
) : __assert_fail ("SplitPos->getParent() == ExitBB && \"Unexpected Insertion point location!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3629, __extension__
 __PRETTY_FUNCTION__))
       "Unexpected Insertion point location!")(static_cast <bool> (SplitPos->getParent() == ExitBB
 && "Unexpected Insertion point location!") ? void (0
) : __assert_fail ("SplitPos->getParent() == ExitBB && \"Unexpected Insertion point location!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3629, __extension__
 __PRETTY_FUNCTION__));
auto merged = MergeBlockIntoPredecessor(ExitBB);
BasicBlock *ExitPredBB = SplitPos->getParent();
auto InsertBB = merged ? ExitPredBB : ExitBB;
if (!isa_and_nonnull<BranchInst>(SplitPos))
  SplitPos->eraseFromParent();
Builder.SetInsertPoint(InsertBB);

return Builder.saveIP();
3638}

3640OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::emitCommonDirectiveEntry(
  Directive OMPD, Value *EntryCall, BasicBlock *ExitBB, bool Conditional) {
// if nothing to do, Return current insertion point.
if (!Conditional || !EntryCall)
  return Builder.saveIP();

BasicBlock *EntryBB = Builder.GetInsertBlock();
Value *CallBool = Builder.CreateIsNotNull(EntryCall);
auto *ThenBB = BasicBlock::Create(M.getContext(), "omp_region.body");
auto *UI = new UnreachableInst(Builder.getContext(), ThenBB);

// Emit thenBB and set the Builder's insertion point there for
// body generation next. Place the block after the current block.
Function *CurFn = EntryBB->getParent();
CurFn->insert(std::next(EntryBB->getIterator()), ThenBB);

// Move Entry branch to end of ThenBB, and replace with conditional
// branch (If-stmt)
Instruction *EntryBBTI = EntryBB->getTerminator();
Builder.CreateCondBr(CallBool, ThenBB, ExitBB);
EntryBBTI->removeFromParent();
Builder.SetInsertPoint(UI);
Builder.Insert(EntryBBTI);
UI->eraseFromParent();
Builder.SetInsertPoint(ThenBB->getTerminator());

// return an insertion point to ExitBB.
return IRBuilder<>::InsertPoint(ExitBB, ExitBB->getFirstInsertionPt());
3668}

3670OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::emitCommonDirectiveExit(
  omp::Directive OMPD, InsertPointTy FinIP, Instruction *ExitCall,
  bool HasFinalize) {

Builder.restoreIP(FinIP);

// If there is finalization to do, emit it before the exit call
if (HasFinalize) {
  assert(!FinalizationStack.empty() &&(static_cast <bool> (!FinalizationStack.empty() &&
 "Unexpected finalization stack state!") ? void (0) : __assert_fail
 ("!FinalizationStack.empty() && \"Unexpected finalization stack state!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3679, __extension__
 __PRETTY_FUNCTION__))
         "Unexpected finalization stack state!")(static_cast <bool> (!FinalizationStack.empty() &&
 "Unexpected finalization stack state!") ? void (0) : __assert_fail
 ("!FinalizationStack.empty() && \"Unexpected finalization stack state!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3679, __extension__
 __PRETTY_FUNCTION__));

  FinalizationInfo Fi = FinalizationStack.pop_back_val();
  assert(Fi.DK == OMPD && "Unexpected Directive for Finalization call!")(static_cast <bool> (Fi.DK == OMPD && "Unexpected Directive for Finalization call!"
) ? void (0) : __assert_fail ("Fi.DK == OMPD && \"Unexpected Directive for Finalization call!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3682, __extension__
 __PRETTY_FUNCTION__));

  Fi.FiniCB(FinIP);

  BasicBlock *FiniBB = FinIP.getBlock();
  Instruction *FiniBBTI = FiniBB->getTerminator();

  // set Builder IP for call creation
  Builder.SetInsertPoint(FiniBBTI);
}

if (!ExitCall)
  return Builder.saveIP();

// place the Exitcall as last instruction before Finalization block terminator
ExitCall->removeFromParent();
Builder.Insert(ExitCall);

return IRBuilder<>::InsertPoint(ExitCall->getParent(),
                                ExitCall->getIterator());
3702}

3704OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::createCopyinClauseBlocks(
  InsertPointTy IP, Value *MasterAddr, Value *PrivateAddr,
  llvm::IntegerType *IntPtrTy, bool BranchtoEnd) {
if (!IP.isSet())
  return IP;

IRBuilder<>::InsertPointGuard IPG(Builder);

// creates the following CFG structure
//	   OMP_Entry : (MasterAddr != PrivateAddr)?
//       F     T
//       |      \
//       |     copin.not.master
//       |      /
//       v     /
//   copyin.not.master.end
//		     |
//         v
//   OMP.Entry.Next

BasicBlock *OMP_Entry = IP.getBlock();
Function *CurFn = OMP_Entry->getParent();
BasicBlock *CopyBegin =
    BasicBlock::Create(M.getContext(), "copyin.not.master", CurFn);
BasicBlock *CopyEnd = nullptr;

// If entry block is terminated, split to preserve the branch to following
// basic block (i.e. OMP.Entry.Next), otherwise, leave everything as is.
if (isa_and_nonnull<BranchInst>(OMP_Entry->getTerminator())) {
  CopyEnd = OMP_Entry->splitBasicBlock(OMP_Entry->getTerminator(),
                                       "copyin.not.master.end");
  OMP_Entry->getTerminator()->eraseFromParent();
} else {
  CopyEnd =
      BasicBlock::Create(M.getContext(), "copyin.not.master.end", CurFn);
}

Builder.SetInsertPoint(OMP_Entry);
Value *MasterPtr = Builder.CreatePtrToInt(MasterAddr, IntPtrTy);
Value *PrivatePtr = Builder.CreatePtrToInt(PrivateAddr, IntPtrTy);
Value *cmp = Builder.CreateICmpNE(MasterPtr, PrivatePtr);
Builder.CreateCondBr(cmp, CopyBegin, CopyEnd);

Builder.SetInsertPoint(CopyBegin);
if (BranchtoEnd)
  Builder.SetInsertPoint(Builder.CreateBr(CopyEnd));

return Builder.saveIP();
3752}

3754CallInst *OpenMPIRBuilder::createOMPAlloc(const LocationDescription &Loc,
                                        Value *Size, Value *Allocator,
                                        std::string Name) {
IRBuilder<>::InsertPointGuard IPG(Builder);
Builder.restoreIP(Loc.IP);

uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
Value *ThreadId = getOrCreateThreadID(Ident);
Value *Args[] = {ThreadId, Size, Allocator};

Function *Fn = getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_alloc);

return Builder.CreateCall(Fn, Args, Name);
3769}

3771CallInst *OpenMPIRBuilder::createOMPFree(const LocationDescription &Loc,
                                       Value *Addr, Value *Allocator,
                                       std::string Name) {
IRBuilder<>::InsertPointGuard IPG(Builder);
Builder.restoreIP(Loc.IP);

uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
Value *ThreadId = getOrCreateThreadID(Ident);
Value *Args[] = {ThreadId, Addr, Allocator};
Function *Fn = getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_free);
return Builder.CreateCall(Fn, Args, Name);
3784}

3786CallInst *OpenMPIRBuilder::createOMPInteropInit(
  const LocationDescription &Loc, Value *InteropVar,
  omp::OMPInteropType InteropType, Value *Device, Value *NumDependences,
  Value *DependenceAddress, bool HaveNowaitClause) {
IRBuilder<>::InsertPointGuard IPG(Builder);
Builder.restoreIP(Loc.IP);

uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
Value *ThreadId = getOrCreateThreadID(Ident);
if (Device == nullptr)
  Device = ConstantInt::get(Int32, -1);
Constant *InteropTypeVal = ConstantInt::get(Int32, (int)InteropType);
if (NumDependences == nullptr) {
  NumDependences = ConstantInt::get(Int32, 0);
  PointerType *PointerTypeVar = Type::getInt8PtrTy(M.getContext());
  DependenceAddress = ConstantPointerNull::get(PointerTypeVar);
}
Value *HaveNowaitClauseVal = ConstantInt::get(Int32, HaveNowaitClause);
Value *Args[] = {
    Ident,  ThreadId,       InteropVar,        InteropTypeVal,
    Device, NumDependences, DependenceAddress, HaveNowaitClauseVal};

Function *Fn = getOrCreateRuntimeFunctionPtr(OMPRTL___tgt_interop_init);

return Builder.CreateCall(Fn, Args);
3813}

3815CallInst *OpenMPIRBuilder::createOMPInteropDestroy(
  const LocationDescription &Loc, Value *InteropVar, Value *Device,
  Value *NumDependences, Value *DependenceAddress, bool HaveNowaitClause) {
IRBuilder<>::InsertPointGuard IPG(Builder);
Builder.restoreIP(Loc.IP);

uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
Value *ThreadId = getOrCreateThreadID(Ident);
if (Device == nullptr)
  Device = ConstantInt::get(Int32, -1);
if (NumDependences == nullptr) {
  NumDependences = ConstantInt::get(Int32, 0);
  PointerType *PointerTypeVar = Type::getInt8PtrTy(M.getContext());
  DependenceAddress = ConstantPointerNull::get(PointerTypeVar);
}
Value *HaveNowaitClauseVal = ConstantInt::get(Int32, HaveNowaitClause);
Value *Args[] = {
    Ident,          ThreadId,          InteropVar,         Device,
    NumDependences, DependenceAddress, HaveNowaitClauseVal};

Function *Fn = getOrCreateRuntimeFunctionPtr(OMPRTL___tgt_interop_destroy);

return Builder.CreateCall(Fn, Args);
3840}

3842CallInst *OpenMPIRBuilder::createOMPInteropUse(const LocationDescription &Loc,
                                             Value *InteropVar, Value *Device,
                                             Value *NumDependences,
                                             Value *DependenceAddress,
                                             bool HaveNowaitClause) {
IRBuilder<>::InsertPointGuard IPG(Builder);
Builder.restoreIP(Loc.IP);
uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
Value *ThreadId = getOrCreateThreadID(Ident);
if (Device == nullptr)
  Device = ConstantInt::get(Int32, -1);
if (NumDependences == nullptr) {
  NumDependences = ConstantInt::get(Int32, 0);
  PointerType *PointerTypeVar = Type::getInt8PtrTy(M.getContext());
  DependenceAddress = ConstantPointerNull::get(PointerTypeVar);
}
Value *HaveNowaitClauseVal = ConstantInt::get(Int32, HaveNowaitClause);
Value *Args[] = {
    Ident,          ThreadId,          InteropVar,         Device,
    NumDependences, DependenceAddress, HaveNowaitClauseVal};

Function *Fn = getOrCreateRuntimeFunctionPtr(OMPRTL___tgt_interop_use);

return Builder.CreateCall(Fn, Args);
3868}

3870CallInst *OpenMPIRBuilder::createCachedThreadPrivate(
  const LocationDescription &Loc, llvm::Value *Pointer,
  llvm::ConstantInt *Size, const llvm::Twine &Name) {
IRBuilder<>::InsertPointGuard IPG(Builder);
Builder.restoreIP(Loc.IP);

uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
Value *ThreadId = getOrCreateThreadID(Ident);
Constant *ThreadPrivateCache =
    getOrCreateInternalVariable(Int8PtrPtr, Name.str());
llvm::Value *Args[] = {Ident, ThreadId, Pointer, Size, ThreadPrivateCache};

Function *Fn =
    getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_threadprivate_cached);

return Builder.CreateCall(Fn, Args);
3888}

3890OpenMPIRBuilder::InsertPointTy
3891OpenMPIRBuilder::createTargetInit(const LocationDescription &Loc, bool IsSPMD) {
if (!updateToLocation(Loc))
  return Loc.IP;

uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Constant *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
ConstantInt *IsSPMDVal = ConstantInt::getSigned(
    IntegerType::getInt8Ty(Int8->getContext()),
    IsSPMD ? OMP_TGT_EXEC_MODE_SPMD : OMP_TGT_EXEC_MODE_GENERIC);
ConstantInt *UseGenericStateMachine =
    ConstantInt::getBool(Int32->getContext(), !IsSPMD);

Function *Fn = getOrCreateRuntimeFunctionPtr(
    omp::RuntimeFunction::OMPRTL___kmpc_target_init);

CallInst *ThreadKind = Builder.CreateCall(
    Fn, {Ident, IsSPMDVal, UseGenericStateMachine});

Value *ExecUserCode = Builder.CreateICmpEQ(
    ThreadKind, ConstantInt::get(ThreadKind->getType(), -1),
    "exec_user_code");

// ThreadKind = __kmpc_target_init(...)
// if (ThreadKind == -1)
//   user_code
// else
//   return;

auto *UI = Builder.CreateUnreachable();
BasicBlock *CheckBB = UI->getParent();
BasicBlock *UserCodeEntryBB = CheckBB->splitBasicBlock(UI, "user_code.entry");

BasicBlock *WorkerExitBB = BasicBlock::Create(
    CheckBB->getContext(), "worker.exit", CheckBB->getParent());
Builder.SetInsertPoint(WorkerExitBB);
Builder.CreateRetVoid();

auto *CheckBBTI = CheckBB->getTerminator();
Builder.SetInsertPoint(CheckBBTI);
Builder.CreateCondBr(ExecUserCode, UI->getParent(), WorkerExitBB);

CheckBBTI->eraseFromParent();
UI->eraseFromParent();

// Continue in the "user_code" block, see diagram above and in
// openmp/libomptarget/deviceRTLs/common/include/target.h .
return InsertPointTy(UserCodeEntryBB, UserCodeEntryBB->getFirstInsertionPt());
3939}

3941void OpenMPIRBuilder::createTargetDeinit(const LocationDescription &Loc,
                                       bool IsSPMD) {
if (!updateToLocation(Loc))
  return;

uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
ConstantInt *IsSPMDVal = ConstantInt::getSigned(
    IntegerType::getInt8Ty(Int8->getContext()),
    IsSPMD ? OMP_TGT_EXEC_MODE_SPMD : OMP_TGT_EXEC_MODE_GENERIC);

Function *Fn = getOrCreateRuntimeFunctionPtr(
    omp::RuntimeFunction::OMPRTL___kmpc_target_deinit);

Builder.CreateCall(Fn, {Ident, IsSPMDVal});
3957}

3959void OpenMPIRBuilder::setOutlinedTargetRegionFunctionAttributes(
  Function *OutlinedFn, int32_t NumTeams, int32_t NumThreads) {
if (Config.isEmbedded()) {
  OutlinedFn->setLinkage(GlobalValue::WeakODRLinkage);
  // TODO: Determine if DSO local can be set to true.
  OutlinedFn->setDSOLocal(false);
  OutlinedFn->setVisibility(GlobalValue::ProtectedVisibility);
  if (Triple(M.getTargetTriple()).isAMDGCN())
    OutlinedFn->setCallingConv(CallingConv::AMDGPU_KERNEL);
}

if (NumTeams > 0)
  OutlinedFn->addFnAttr("omp_target_num_teams", std::to_string(NumTeams));
if (NumThreads > 0)
  OutlinedFn->addFnAttr("omp_target_thread_limit",
                        std::to_string(NumThreads));
3975}

3977Constant *OpenMPIRBuilder::createOutlinedFunctionID(Function *OutlinedFn,
                                                  StringRef EntryFnIDName) {
if (Config.isEmbedded()) {
  assert(OutlinedFn && "The outlined function must exist if embedded")(static_cast <bool> (OutlinedFn && "The outlined function must exist if embedded"
) ? void (0) : __assert_fail ("OutlinedFn && \"The outlined function must exist if embedded\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3980, __extension__
 __PRETTY_FUNCTION__));
  return ConstantExpr::getBitCast(OutlinedFn, Builder.getInt8PtrTy());
}

return new GlobalVariable(
    M, Builder.getInt8Ty(), /*isConstant=*/true, GlobalValue::WeakAnyLinkage,
    Constant::getNullValue(Builder.getInt8Ty()), EntryFnIDName);
3987}

3989Constant *OpenMPIRBuilder::createTargetRegionEntryAddr(Function *OutlinedFn,
                                                     StringRef EntryFnName) {
if (OutlinedFn)
  return OutlinedFn;

assert(!M.getGlobalVariable(EntryFnName, true) &&(static_cast <bool> (!M.getGlobalVariable(EntryFnName, true
) && "Named kernel already exists?") ? void (0) : __assert_fail
 ("!M.getGlobalVariable(EntryFnName, true) && \"Named kernel already exists?\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3995, __extension__
 __PRETTY_FUNCTION__))
       "Named kernel already exists?")(static_cast <bool> (!M.getGlobalVariable(EntryFnName, true
) && "Named kernel already exists?") ? void (0) : __assert_fail
 ("!M.getGlobalVariable(EntryFnName, true) && \"Named kernel already exists?\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 3995, __extension__
 __PRETTY_FUNCTION__));
return new GlobalVariable(
    M, Builder.getInt8Ty(), /*isConstant=*/true, GlobalValue::InternalLinkage,
    Constant::getNullValue(Builder.getInt8Ty()), EntryFnName);
3999}

4001void OpenMPIRBuilder::emitTargetRegionFunction(
  TargetRegionEntryInfo &EntryInfo,
  FunctionGenCallback &GenerateFunctionCallback, int32_t NumTeams,
  int32_t NumThreads, bool IsOffloadEntry, Function *&OutlinedFn,
  Constant *&OutlinedFnID) {

SmallString<64> EntryFnName;
OffloadInfoManager.getTargetRegionEntryFnName(EntryFnName, EntryInfo);

OutlinedFn = Config.isEmbedded() || !Config.openMPOffloadMandatory()
                 ? GenerateFunctionCallback(EntryFnName)
                 : nullptr;

// If this target outline function is not an offload entry, we don't need to
// register it. This may be in the case of a false if clause, or if there are
// no OpenMP targets.
if (!IsOffloadEntry)
  return;

std::string EntryFnIDName =
    Config.isEmbedded()
        ? std::string(EntryFnName)
        : createPlatformSpecificName({EntryFnName, "region_id"});

OutlinedFnID = registerTargetRegionFunction(
    EntryInfo, OutlinedFn, EntryFnName, EntryFnIDName, NumTeams, NumThreads);
4027}

4029Constant *OpenMPIRBuilder::registerTargetRegionFunction(
  TargetRegionEntryInfo &EntryInfo, Function *OutlinedFn,
  StringRef EntryFnName, StringRef EntryFnIDName, int32_t NumTeams,
  int32_t NumThreads) {
if (OutlinedFn)
  setOutlinedTargetRegionFunctionAttributes(OutlinedFn, NumTeams, NumThreads);
auto OutlinedFnID = createOutlinedFunctionID(OutlinedFn, EntryFnIDName);
auto EntryAddr = createTargetRegionEntryAddr(OutlinedFn, EntryFnName);
OffloadInfoManager.registerTargetRegionEntryInfo(
    EntryInfo, EntryAddr, OutlinedFnID,
    OffloadEntriesInfoManager::OMPTargetRegionEntryTargetRegion);
return OutlinedFnID;
4041}

4043OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::createTargetData(
  const LocationDescription &Loc, OpenMPIRBuilder::InsertPointTy CodeGenIP,
  SmallVectorImpl<uint64_t> &MapTypeFlags,
  SmallVectorImpl<Constant *> &MapNames, struct MapperAllocas &MapperAllocas,
  bool IsBegin, int64_t DeviceID, Value *IfCond,
  BodyGenCallbackTy ProcessMapOpCB, BodyGenCallbackTy BodyGenCB) {
if (!updateToLocation(Loc))
  return InsertPointTy();

Builder.restoreIP(CodeGenIP);

// LLVM utilities like blocks with terminators.
// The UI acts as a resume point for code insertion after the BodyGen
auto *UI = Builder.CreateUnreachable();
if (IfCond) {
  auto *ThenTI =
      SplitBlockAndInsertIfThen(IfCond, UI, /* Unreachable */ false);
  ThenTI->getParent()->setName("omp_if.then");
  Builder.SetInsertPoint(ThenTI);
} else {
  Builder.SetInsertPoint(UI);
}

ProcessMapOpCB(Builder.saveIP(), Builder.saveIP());

uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *srcLocInfo = getOrCreateIdent(SrcLocStr, SrcLocStrSize);

GlobalVariable *MapTypesGV =
    createOffloadMaptypes(MapTypeFlags, ".offload_maptypes");
Value *MapTypesArg = Builder.CreateConstInBoundsGEP2_32(
    ArrayType::get(Builder.getInt64Ty(), MapTypeFlags.size()), MapTypesGV,
    /*Idx0=*/0, /*Idx1=*/0);

GlobalVariable *MapNamesGV =
    createOffloadMapnames(MapNames, ".offload_mapnames");
Value *MapNamesArg = Builder.CreateConstInBoundsGEP2_32(
    ArrayType::get(Builder.getInt8PtrTy(), MapNames.size()), MapNamesGV,
    /*Idx0=*/0, /*Idx1=*/0);

Function *beginMapperFunc =
    getOrCreateRuntimeFunctionPtr(omp::OMPRTL___tgt_target_data_begin_mapper);
Function *endMapperFunc =
    getOrCreateRuntimeFunctionPtr(omp::OMPRTL___tgt_target_data_end_mapper);

if (BodyGenCB) {
  // Create call to start the data region.
  emitMapperCall(Builder.saveIP(), beginMapperFunc, srcLocInfo, MapTypesArg,
                 MapNamesArg, MapperAllocas, DeviceID, MapTypeFlags.size());

  BodyGenCB(Builder.saveIP(), Builder.saveIP());

  Builder.SetInsertPoint(UI->getParent());
  // Create call to end the data region.
  emitMapperCall(Builder.saveIP(), endMapperFunc, srcLocInfo, MapTypesArg,
                 MapNamesArg, MapperAllocas, DeviceID, MapTypeFlags.size());
} else {
  emitMapperCall(Builder.saveIP(), IsBegin ? beginMapperFunc : endMapperFunc,
                 srcLocInfo, MapTypesArg, MapNamesArg, MapperAllocas,
                 DeviceID, MapTypeFlags.size());
}

// Update the insertion point and remove the terminator we introduced.
Builder.SetInsertPoint(UI->getParent());
if (IfCond)
  UI->getParent()->setName("omp_if.end");
UI->eraseFromParent();
return Builder.saveIP();
4112}

4114static Function *
4115createOutlinedFunction(IRBuilderBase &Builder, StringRef FuncName,
                     SmallVectorImpl<Value *> &Inputs,
                     OpenMPIRBuilder::TargetBodyGenCallbackTy &CBFunc) {
SmallVector<Type *> ParameterTypes;
for (auto &Arg : Inputs)
  ParameterTypes.push_back(Arg->getType());

auto FuncType = FunctionType::get(Builder.getVoidTy(), ParameterTypes,
                                  /*isVarArg*/ false);
auto Func = Function::Create(FuncType, GlobalValue::InternalLinkage, FuncName,
                             Builder.GetInsertBlock()->getModule());

// Save insert point.
auto OldInsertPoint = Builder.saveIP();

// Generate the region into the function.
BasicBlock *EntryBB = BasicBlock::Create(Builder.getContext(), "entry", Func);
Builder.SetInsertPoint(EntryBB);
Builder.restoreIP(CBFunc(Builder.saveIP(), Builder.saveIP()));

// Insert return instruction.
Builder.CreateRetVoid();

// Rewrite uses of input valus to parameters.
for (auto InArg : zip(Inputs, Func->args())) {
  Value *Input = std::get<0>(InArg);
  Argument &Arg = std::get<1>(InArg);

  // Collect all the instructions
  for (User *User : make_early_inc_range(Input->users()))
    if (auto Instr = dyn_cast<Instruction>(User))
      if (Instr->getFunction() == Func)
        Instr->replaceUsesOfWith(Input, &Arg);
}

// Restore insert point.
Builder.restoreIP(OldInsertPoint);

return Func;
4154}

4156static void
4157emitTargetOutlinedFunction(OpenMPIRBuilder &OMPBuilder, IRBuilderBase &Builder,
                         TargetRegionEntryInfo &EntryInfo,
                         Function *&OutlinedFn, int32_t NumTeams,
                         int32_t NumThreads, SmallVectorImpl<Value *> &Inputs,
                         OpenMPIRBuilder::TargetBodyGenCallbackTy &CBFunc) {

OpenMPIRBuilder::FunctionGenCallback &&GenerateOutlinedFunction =
    [&Builder, &Inputs, &CBFunc](StringRef EntryFnName) {
      return createOutlinedFunction(Builder, EntryFnName, Inputs, CBFunc);
    };

Constant *OutlinedFnID;
OMPBuilder.emitTargetRegionFunction(EntryInfo, GenerateOutlinedFunction,
                                    NumTeams, NumThreads, true, OutlinedFn,
                                    OutlinedFnID);
4172}

4174static void emitTargetCall(IRBuilderBase &Builder, Function *OutlinedFn,
                         SmallVectorImpl<Value *> &Args) {
// TODO: Add kernel launch call when device codegen is supported.
Builder.CreateCall(OutlinedFn, Args);
4178}

4180OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::createTarget(
  const LocationDescription &Loc, OpenMPIRBuilder::InsertPointTy CodeGenIP,
  TargetRegionEntryInfo &EntryInfo, int32_t NumTeams, int32_t NumThreads,
  SmallVectorImpl<Value *> &Args, TargetBodyGenCallbackTy CBFunc) {
if (!updateToLocation(Loc))
  return InsertPointTy();

Builder.restoreIP(CodeGenIP);

Function *OutlinedFn;
emitTargetOutlinedFunction(*this, Builder, EntryInfo, OutlinedFn, NumTeams,
                           NumThreads, Args, CBFunc);
emitTargetCall(Builder, OutlinedFn, Args);
return Builder.saveIP();
4194}

4196std::string OpenMPIRBuilder::getNameWithSeparators(ArrayRef<StringRef> Parts,
                                                 StringRef FirstSeparator,
                                                 StringRef Separator) {
SmallString<128> Buffer;
llvm::raw_svector_ostream OS(Buffer);
StringRef Sep = FirstSeparator;
for (StringRef Part : Parts) {
  OS << Sep << Part;
  Sep = Separator;
}
return OS.str().str();
4207}

4209std::string
4210OpenMPIRBuilder::createPlatformSpecificName(ArrayRef<StringRef> Parts) const {
return OpenMPIRBuilder::getNameWithSeparators(Parts, Config.firstSeparator(),
                                              Config.separator());
4213}

4215GlobalVariable *
4216OpenMPIRBuilder::getOrCreateInternalVariable(Type *Ty, const StringRef &Name,
                                           unsigned AddressSpace) {
auto &Elem = *InternalVars.try_emplace(Name, nullptr).first;
if (Elem.second) {
  assert(cast<PointerType>(Elem.second->getType())(static_cast <bool> (cast<PointerType>(Elem.second
->getType()) ->isOpaqueOrPointeeTypeMatches(Ty) &&
 "OMP internal variable has different type than requested") ?
 void (0) : __assert_fail ("cast<PointerType>(Elem.second->getType()) ->isOpaqueOrPointeeTypeMatches(Ty) && \"OMP internal variable has different type than requested\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4222, __extension__
 __PRETTY_FUNCTION__))
             ->isOpaqueOrPointeeTypeMatches(Ty) &&(static_cast <bool> (cast<PointerType>(Elem.second
->getType()) ->isOpaqueOrPointeeTypeMatches(Ty) &&
 "OMP internal variable has different type than requested") ?
 void (0) : __assert_fail ("cast<PointerType>(Elem.second->getType()) ->isOpaqueOrPointeeTypeMatches(Ty) && \"OMP internal variable has different type than requested\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4222, __extension__
 __PRETTY_FUNCTION__))
         "OMP internal variable has different type than requested")(static_cast <bool> (cast<PointerType>(Elem.second
->getType()) ->isOpaqueOrPointeeTypeMatches(Ty) &&
 "OMP internal variable has different type than requested") ?
 void (0) : __assert_fail ("cast<PointerType>(Elem.second->getType()) ->isOpaqueOrPointeeTypeMatches(Ty) && \"OMP internal variable has different type than requested\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4222, __extension__
 __PRETTY_FUNCTION__));
} else {
  // TODO: investigate the appropriate linkage type used for the global
  // variable for possibly changing that to internal or private, or maybe
  // create different versions of the function for different OMP internal
  // variables.
  Elem.second = new GlobalVariable(
      M, Ty, /*IsConstant=*/false, GlobalValue::CommonLinkage,
      Constant::getNullValue(Ty), Elem.first(),
      /*InsertBefore=*/nullptr, GlobalValue::NotThreadLocal, AddressSpace);
}

return cast<GlobalVariable>(&*Elem.second);
4235}

4237Value *OpenMPIRBuilder::getOMPCriticalRegionLock(StringRef CriticalName) {
std::string Prefix = Twine("gomp_critical_user_", CriticalName).str();
std::string Name = getNameWithSeparators({Prefix, "var"}, ".", ".");
return getOrCreateInternalVariable(KmpCriticalNameTy, Name);
4241}

4243Value *OpenMPIRBuilder::getSizeInBytes(Value *BasePtr) {
LLVMContext &Ctx = Builder.getContext();
Value *Null = Constant::getNullValue(BasePtr->getType()->getPointerTo());
Value *SizeGep =
    Builder.CreateGEP(BasePtr->getType(), Null, Builder.getInt32(1));
Value *SizePtrToInt = Builder.CreatePtrToInt(SizeGep, Type::getInt64Ty(Ctx));
return SizePtrToInt;
4250}

4252GlobalVariable *
4253OpenMPIRBuilder::createOffloadMaptypes(SmallVectorImpl<uint64_t> &Mappings,
                                     std::string VarName) {
llvm::Constant *MaptypesArrayInit =
    llvm::ConstantDataArray::get(M.getContext(), Mappings);
auto *MaptypesArrayGlobal = new llvm::GlobalVariable(
    M, MaptypesArrayInit->getType(),
    /*isConstant=*/true, llvm::GlobalValue::PrivateLinkage, MaptypesArrayInit,
    VarName);
MaptypesArrayGlobal->setUnnamedAddr(llvm::GlobalValue::UnnamedAddr::Global);
return MaptypesArrayGlobal;
4263}

4265void OpenMPIRBuilder::createMapperAllocas(const LocationDescription &Loc,
                                        InsertPointTy AllocaIP,
                                        unsigned NumOperands,
                                        struct MapperAllocas &MapperAllocas) {
if (!updateToLocation(Loc))
  return;

auto *ArrI8PtrTy = ArrayType::get(Int8Ptr, NumOperands);
auto *ArrI64Ty = ArrayType::get(Int64, NumOperands);
Builder.restoreIP(AllocaIP);
AllocaInst *ArgsBase = Builder.CreateAlloca(
    ArrI8PtrTy, /* ArraySize = */ nullptr, ".offload_baseptrs");
AllocaInst *Args = Builder.CreateAlloca(ArrI8PtrTy, /* ArraySize = */ nullptr,
                                        ".offload_ptrs");
AllocaInst *ArgSizes = Builder.CreateAlloca(
    ArrI64Ty, /* ArraySize = */ nullptr, ".offload_sizes");
Builder.restoreIP(Loc.IP);
MapperAllocas.ArgsBase = ArgsBase;
MapperAllocas.Args = Args;
MapperAllocas.ArgSizes = ArgSizes;
4285}

4287void OpenMPIRBuilder::emitMapperCall(const LocationDescription &Loc,
                                   Function *MapperFunc, Value *SrcLocInfo,
                                   Value *MaptypesArg, Value *MapnamesArg,
                                   struct MapperAllocas &MapperAllocas,
                                   int64_t DeviceID, unsigned NumOperands) {
if (!updateToLocation(Loc))
  return;

auto *ArrI8PtrTy = ArrayType::get(Int8Ptr, NumOperands);
auto *ArrI64Ty = ArrayType::get(Int64, NumOperands);
Value *ArgsBaseGEP =
    Builder.CreateInBoundsGEP(ArrI8PtrTy, MapperAllocas.ArgsBase,
                              {Builder.getInt32(0), Builder.getInt32(0)});
Value *ArgsGEP =
    Builder.CreateInBoundsGEP(ArrI8PtrTy, MapperAllocas.Args,
                              {Builder.getInt32(0), Builder.getInt32(0)});
Value *ArgSizesGEP =
    Builder.CreateInBoundsGEP(ArrI64Ty, MapperAllocas.ArgSizes,
                              {Builder.getInt32(0), Builder.getInt32(0)});
Value *NullPtr = Constant::getNullValue(Int8Ptr->getPointerTo());
Builder.CreateCall(MapperFunc,
                   {SrcLocInfo, Builder.getInt64(DeviceID),
                    Builder.getInt32(NumOperands), ArgsBaseGEP, ArgsGEP,
                    ArgSizesGEP, MaptypesArg, MapnamesArg, NullPtr});
4311}

4313void OpenMPIRBuilder::emitOffloadingArraysArgument(IRBuilderBase &Builder,
                                                 TargetDataRTArgs &RTArgs,
                                                 TargetDataInfo &Info,
                                                 bool EmitDebug,
                                                 bool ForEndCall) {
assert((!ForEndCall || Info.separateBeginEndCalls()) &&(static_cast <bool> ((!ForEndCall || Info.separateBeginEndCalls
()) && "expected region end call to runtime only when end call is separate"
) ? void (0) : __assert_fail ("(!ForEndCall || Info.separateBeginEndCalls()) && \"expected region end call to runtime only when end call is separate\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4319, __extension__
 __PRETTY_FUNCTION__))
       "expected region end call to runtime only when end call is separate")(static_cast <bool> ((!ForEndCall || Info.separateBeginEndCalls
()) && "expected region end call to runtime only when end call is separate"
) ? void (0) : __assert_fail ("(!ForEndCall || Info.separateBeginEndCalls()) && \"expected region end call to runtime only when end call is separate\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4319, __extension__
 __PRETTY_FUNCTION__));
auto VoidPtrTy = Type::getInt8PtrTy(M.getContext());
auto VoidPtrPtrTy = VoidPtrTy->getPointerTo(0);
auto Int64Ty = Type::getInt64Ty(M.getContext());
auto Int64PtrTy = Type::getInt64PtrTy(M.getContext());

if (!Info.NumberOfPtrs) {
  RTArgs.BasePointersArray = ConstantPointerNull::get(VoidPtrPtrTy);
  RTArgs.PointersArray = ConstantPointerNull::get(VoidPtrPtrTy);
  RTArgs.SizesArray = ConstantPointerNull::get(Int64PtrTy);
  RTArgs.MapTypesArray = ConstantPointerNull::get(Int64PtrTy);
  RTArgs.MapNamesArray = ConstantPointerNull::get(VoidPtrPtrTy);
  RTArgs.MappersArray = ConstantPointerNull::get(VoidPtrPtrTy);
  return;
}

RTArgs.BasePointersArray = Builder.CreateConstInBoundsGEP2_32(
    ArrayType::get(VoidPtrTy, Info.NumberOfPtrs),
    Info.RTArgs.BasePointersArray,
    /*Idx0=*/0, /*Idx1=*/0);
RTArgs.PointersArray = Builder.CreateConstInBoundsGEP2_32(
    ArrayType::get(VoidPtrTy, Info.NumberOfPtrs), Info.RTArgs.PointersArray,
    /*Idx0=*/0,
    /*Idx1=*/0);
RTArgs.SizesArray = Builder.CreateConstInBoundsGEP2_32(
    ArrayType::get(Int64Ty, Info.NumberOfPtrs), Info.RTArgs.SizesArray,
    /*Idx0=*/0, /*Idx1=*/0);
RTArgs.MapTypesArray = Builder.CreateConstInBoundsGEP2_32(
    ArrayType::get(Int64Ty, Info.NumberOfPtrs),
    ForEndCall && Info.RTArgs.MapTypesArrayEnd ? Info.RTArgs.MapTypesArrayEnd
                                               : Info.RTArgs.MapTypesArray,
    /*Idx0=*/0,
    /*Idx1=*/0);

// Only emit the mapper information arrays if debug information is
// requested.
if (!EmitDebug)
  RTArgs.MapNamesArray = ConstantPointerNull::get(VoidPtrPtrTy);
else
  RTArgs.MapNamesArray = Builder.CreateConstInBoundsGEP2_32(
      ArrayType::get(VoidPtrTy, Info.NumberOfPtrs), Info.RTArgs.MapNamesArray,
      /*Idx0=*/0,
      /*Idx1=*/0);
// If there is no user-defined mapper, set the mapper array to nullptr to
// avoid an unnecessary data privatization
if (!Info.HasMapper)
  RTArgs.MappersArray = ConstantPointerNull::get(VoidPtrPtrTy);
else
  RTArgs.MappersArray =
      Builder.CreatePointerCast(Info.RTArgs.MappersArray, VoidPtrPtrTy);
4369}

4371bool OpenMPIRBuilder::checkAndEmitFlushAfterAtomic(
  const LocationDescription &Loc, llvm::AtomicOrdering AO, AtomicKind AK) {
assert(!(AO == AtomicOrdering::NotAtomic ||(static_cast <bool> (!(AO == AtomicOrdering::NotAtomic ||
 AO == llvm::AtomicOrdering::Unordered) && "Unexpected Atomic Ordering."
) ? void (0) : __assert_fail ("!(AO == AtomicOrdering::NotAtomic || AO == llvm::AtomicOrdering::Unordered) && \"Unexpected Atomic Ordering.\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4375, __extension__
 __PRETTY_FUNCTION__))
         AO == llvm::AtomicOrdering::Unordered) &&(static_cast <bool> (!(AO == AtomicOrdering::NotAtomic ||
 AO == llvm::AtomicOrdering::Unordered) && "Unexpected Atomic Ordering."
) ? void (0) : __assert_fail ("!(AO == AtomicOrdering::NotAtomic || AO == llvm::AtomicOrdering::Unordered) && \"Unexpected Atomic Ordering.\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4375, __extension__
 __PRETTY_FUNCTION__))
       "Unexpected Atomic Ordering.")(static_cast <bool> (!(AO == AtomicOrdering::NotAtomic ||
 AO == llvm::AtomicOrdering::Unordered) && "Unexpected Atomic Ordering."
) ? void (0) : __assert_fail ("!(AO == AtomicOrdering::NotAtomic || AO == llvm::AtomicOrdering::Unordered) && \"Unexpected Atomic Ordering.\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4375, __extension__
 __PRETTY_FUNCTION__));

bool Flush = false;
llvm::AtomicOrdering FlushAO = AtomicOrdering::Monotonic;

switch (AK) {
case Read:
  if (AO == AtomicOrdering::Acquire || AO == AtomicOrdering::AcquireRelease ||
      AO == AtomicOrdering::SequentiallyConsistent) {
    FlushAO = AtomicOrdering::Acquire;
    Flush = true;
  }
  break;
case Write:
case Compare:
case Update:
  if (AO == AtomicOrdering::Release || AO == AtomicOrdering::AcquireRelease ||
      AO == AtomicOrdering::SequentiallyConsistent) {
    FlushAO = AtomicOrdering::Release;
    Flush = true;
  }
  break;
case Capture:
  switch (AO) {
  case AtomicOrdering::Acquire:
    FlushAO = AtomicOrdering::Acquire;
    Flush = true;
    break;
  case AtomicOrdering::Release:
    FlushAO = AtomicOrdering::Release;
    Flush = true;
    break;
  case AtomicOrdering::AcquireRelease:
  case AtomicOrdering::SequentiallyConsistent:
    FlushAO = AtomicOrdering::AcquireRelease;
    Flush = true;
    break;
  default:
    // do nothing - leave silently.
    break;
  }
}

if (Flush) {
  // Currently Flush RT call still doesn't take memory_ordering, so for when
  // that happens, this tries to do the resolution of which atomic ordering
  // to use with but issue the flush call
  // TODO: pass `FlushAO` after memory ordering support is added
  (void)FlushAO;
  emitFlush(Loc);
}

// for AO == AtomicOrdering::Monotonic and  all other case combinations
// do nothing
return Flush;
4430}

4432OpenMPIRBuilder::InsertPointTy
4433OpenMPIRBuilder::createAtomicRead(const LocationDescription &Loc,
                                AtomicOpValue &X, AtomicOpValue &V,
                                AtomicOrdering AO) {
if (!updateToLocation(Loc))
  return Loc.IP;

Type *XTy = X.Var->getType();
assert(XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory")(static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4440, __extension__
 __PRETTY_FUNCTION__));
Type *XElemTy = X.ElemTy;
assert((XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() ||(static_cast <bool> ((XElemTy->isFloatingPointTy() ||
 XElemTy->isIntegerTy() || XElemTy->isPointerTy()) &&
 "OMP atomic read expected a scalar type") ? void (0) : __assert_fail
 ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic read expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4444, __extension__
 __PRETTY_FUNCTION__))
        XElemTy->isPointerTy()) &&(static_cast <bool> ((XElemTy->isFloatingPointTy() ||
 XElemTy->isIntegerTy() || XElemTy->isPointerTy()) &&
 "OMP atomic read expected a scalar type") ? void (0) : __assert_fail
 ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic read expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4444, __extension__
 __PRETTY_FUNCTION__))
       "OMP atomic read expected a scalar type")(static_cast <bool> ((XElemTy->isFloatingPointTy() ||
 XElemTy->isIntegerTy() || XElemTy->isPointerTy()) &&
 "OMP atomic read expected a scalar type") ? void (0) : __assert_fail
 ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic read expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4444, __extension__
 __PRETTY_FUNCTION__));

Value *XRead = nullptr;

if (XElemTy->isIntegerTy()) {
  LoadInst *XLD =
      Builder.CreateLoad(XElemTy, X.Var, X.IsVolatile, "omp.atomic.read");
  XLD->setAtomic(AO);
  XRead = cast<Value>(XLD);
} else {
  // We need to bitcast and perform atomic op as integer
  unsigned Addrspace = cast<PointerType>(XTy)->getAddressSpace();
  IntegerType *IntCastTy =
      IntegerType::get(M.getContext(), XElemTy->getScalarSizeInBits());
  Value *XBCast = Builder.CreateBitCast(
      X.Var, IntCastTy->getPointerTo(Addrspace), "atomic.src.int.cast");
  LoadInst *XLoad =
      Builder.CreateLoad(IntCastTy, XBCast, X.IsVolatile, "omp.atomic.load");
  XLoad->setAtomic(AO);
  if (XElemTy->isFloatingPointTy()) {
    XRead = Builder.CreateBitCast(XLoad, XElemTy, "atomic.flt.cast");
  } else {
    XRead = Builder.CreateIntToPtr(XLoad, XElemTy, "atomic.ptr.cast");
  }
}
checkAndEmitFlushAfterAtomic(Loc, AO, AtomicKind::Read);
Builder.CreateStore(XRead, V.Var, V.IsVolatile);
return Builder.saveIP();
4472}

4474OpenMPIRBuilder::InsertPointTy
4475OpenMPIRBuilder::createAtomicWrite(const LocationDescription &Loc,
                                 AtomicOpValue &X, Value *Expr,
                                 AtomicOrdering AO) {
if (!updateToLocation(Loc))
  return Loc.IP;

Type *XTy = X.Var->getType();
assert(XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory")(static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4482, __extension__
 __PRETTY_FUNCTION__));
Type *XElemTy = X.ElemTy;
assert((XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() ||(static_cast <bool> ((XElemTy->isFloatingPointTy() ||
 XElemTy->isIntegerTy() || XElemTy->isPointerTy()) &&
 "OMP atomic write expected a scalar type") ? void (0) : __assert_fail
 ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic write expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4486, __extension__
 __PRETTY_FUNCTION__))
        XElemTy->isPointerTy()) &&(static_cast <bool> ((XElemTy->isFloatingPointTy() ||
 XElemTy->isIntegerTy() || XElemTy->isPointerTy()) &&
 "OMP atomic write expected a scalar type") ? void (0) : __assert_fail
 ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic write expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4486, __extension__
 __PRETTY_FUNCTION__))
       "OMP atomic write expected a scalar type")(static_cast <bool> ((XElemTy->isFloatingPointTy() ||
 XElemTy->isIntegerTy() || XElemTy->isPointerTy()) &&
 "OMP atomic write expected a scalar type") ? void (0) : __assert_fail
 ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic write expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4486, __extension__
 __PRETTY_FUNCTION__));

if (XElemTy->isIntegerTy()) {
  StoreInst *XSt = Builder.CreateStore(Expr, X.Var, X.IsVolatile);
  XSt->setAtomic(AO);
} else {
  // We need to bitcast and perform atomic op as integers
  unsigned Addrspace = cast<PointerType>(XTy)->getAddressSpace();
  IntegerType *IntCastTy =
      IntegerType::get(M.getContext(), XElemTy->getScalarSizeInBits());
  Value *XBCast = Builder.CreateBitCast(
      X.Var, IntCastTy->getPointerTo(Addrspace), "atomic.dst.int.cast");
  Value *ExprCast =
      Builder.CreateBitCast(Expr, IntCastTy, "atomic.src.int.cast");
  StoreInst *XSt = Builder.CreateStore(ExprCast, XBCast, X.IsVolatile);
  XSt->setAtomic(AO);
}

checkAndEmitFlushAfterAtomic(Loc, AO, AtomicKind::Write);
return Builder.saveIP();
4506}

4508OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::createAtomicUpdate(
  const LocationDescription &Loc, InsertPointTy AllocaIP, AtomicOpValue &X,
  Value *Expr, AtomicOrdering AO, AtomicRMWInst::BinOp RMWOp,
  AtomicUpdateCallbackTy &UpdateOp, bool IsXBinopExpr) {
assert(!isConflictIP(Loc.IP, AllocaIP) && "IPs must not be ambiguous")(static_cast <bool> (!isConflictIP(Loc.IP, AllocaIP) &&
 "IPs must not be ambiguous") ? void (0) : __assert_fail ("!isConflictIP(Loc.IP, AllocaIP) && \"IPs must not be ambiguous\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4512, __extension__
 __PRETTY_FUNCTION__));
if (!updateToLocation(Loc))
  return Loc.IP;

LLVM_DEBUG({do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4519, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic update expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic update expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4523, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp
 != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::
UMin) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::UMin) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4526, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
  Type *XTy = X.Var->getType();do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4519, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic update expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic update expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4523, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp
 != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::
UMin) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::UMin) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4526, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
  assert(XTy->isPointerTy() &&do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4519, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic update expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic update expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4523, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp
 != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::
UMin) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::UMin) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4526, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
         "OMP Atomic expects a pointer to target memory");do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4519, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic update expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic update expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4523, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp
 != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::
UMin) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::UMin) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4526, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
  Type *XElemTy = X.ElemTy;do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4519, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic update expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic update expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4523, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp
 != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::
UMin) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::UMin) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4526, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
  assert((XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() ||do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4519, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic update expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic update expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4523, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp
 != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::
UMin) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::UMin) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4526, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
          XElemTy->isPointerTy()) &&do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4519, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic update expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic update expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4523, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp
 != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::
UMin) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::UMin) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4526, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
         "OMP atomic update expected a scalar type");do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4519, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic update expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic update expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4523, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp
 != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::
UMin) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::UMin) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4526, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
  assert((RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) &&do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4519, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic update expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic update expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4523, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp
 != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::
UMin) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::UMin) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4526, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
         (RMWOp != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::UMin) &&do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4519, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic update expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic update expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4523, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp
 != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::
UMin) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::UMin) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4526, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
         "OpenMP atomic does not support LT or GT operations");do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4519, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic update expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic update expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4523, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp
 != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::
UMin) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::UMin) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4526, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
})do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4519, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic update expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic update expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4523, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp
 != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::
UMin) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && (RMWOp != AtomicRMWInst::UMax) && (RMWOp != AtomicRMWInst::UMin) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4526, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false);

emitAtomicUpdate(AllocaIP, X.Var, X.ElemTy, Expr, AO, RMWOp, UpdateOp,
                 X.IsVolatile, IsXBinopExpr);
checkAndEmitFlushAfterAtomic(Loc, AO, AtomicKind::Update);
return Builder.saveIP();
4533}

4535// FIXME: Duplicating AtomicExpand
4536Value *OpenMPIRBuilder::emitRMWOpAsInstruction(Value *Src1, Value *Src2,
                                             AtomicRMWInst::BinOp RMWOp) {
switch (RMWOp) {
case AtomicRMWInst::Add:
  return Builder.CreateAdd(Src1, Src2);
case AtomicRMWInst::Sub:
  return Builder.CreateSub(Src1, Src2);
case AtomicRMWInst::And:
  return Builder.CreateAnd(Src1, Src2);
case AtomicRMWInst::Nand:
  return Builder.CreateNeg(Builder.CreateAnd(Src1, Src2));
case AtomicRMWInst::Or:
  return Builder.CreateOr(Src1, Src2);
case AtomicRMWInst::Xor:
  return Builder.CreateXor(Src1, Src2);
case AtomicRMWInst::Xchg:
case AtomicRMWInst::FAdd:
case AtomicRMWInst::FSub:
case AtomicRMWInst::BAD_BINOP:
case AtomicRMWInst::Max:
case AtomicRMWInst::Min:
case AtomicRMWInst::UMax:
case AtomicRMWInst::UMin:
case AtomicRMWInst::FMax:
case AtomicRMWInst::FMin:
case AtomicRMWInst::UIncWrap:
case AtomicRMWInst::UDecWrap:
  llvm_unreachable("Unsupported atomic update operation")::llvm::llvm_unreachable_internal("Unsupported atomic update operation"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4563);
}
llvm_unreachable("Unsupported atomic update operation")::llvm::llvm_unreachable_internal("Unsupported atomic update operation"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4565);
4566}

4568std::pair<Value *, Value *> OpenMPIRBuilder::emitAtomicUpdate(
  InsertPointTy AllocaIP, Value *X, Type *XElemTy, Value *Expr,
  AtomicOrdering AO, AtomicRMWInst::BinOp RMWOp,
  AtomicUpdateCallbackTy &UpdateOp, bool VolatileX, bool IsXBinopExpr) {
// TODO: handle the case where XElemTy is not byte-sized or not a power of 2
// or a complex datatype.
bool emitRMWOp = false;
switch (RMWOp) {
case AtomicRMWInst::Add:
case AtomicRMWInst::And:
case AtomicRMWInst::Nand:
case AtomicRMWInst::Or:
case AtomicRMWInst::Xor:
case AtomicRMWInst::Xchg:
  emitRMWOp = XElemTy;
  break;
case AtomicRMWInst::Sub:
  emitRMWOp = (IsXBinopExpr && XElemTy);
  break;
default:
  emitRMWOp = false;
}
emitRMWOp &= XElemTy->isIntegerTy();

std::pair<Value *, Value *> Res;
if (emitRMWOp) {
  Res.first = Builder.CreateAtomicRMW(RMWOp, X, Expr, llvm::MaybeAlign(), AO);
  // not needed except in case of postfix captures. Generate anyway for
  // consistency with the else part. Will be removed with any DCE pass.
  // AtomicRMWInst::Xchg does not have a coressponding instruction.
  if (RMWOp == AtomicRMWInst::Xchg)
    Res.second = Res.first;
  else
    Res.second = emitRMWOpAsInstruction(Res.first, Expr, RMWOp);
} else {
  unsigned Addrspace = cast<PointerType>(X->getType())->getAddressSpace();
  IntegerType *IntCastTy =
      IntegerType::get(M.getContext(), XElemTy->getScalarSizeInBits());
  Value *XBCast =
      Builder.CreateBitCast(X, IntCastTy->getPointerTo(Addrspace));
  LoadInst *OldVal =
      Builder.CreateLoad(IntCastTy, XBCast, X->getName() + ".atomic.load");
  OldVal->setAtomic(AO);
  // CurBB
  // |     /---\
// ContBB    |
  // |     \---/
  // ExitBB
  BasicBlock *CurBB = Builder.GetInsertBlock();
  Instruction *CurBBTI = CurBB->getTerminator();
  CurBBTI = CurBBTI ? CurBBTI : Builder.CreateUnreachable();
  BasicBlock *ExitBB =
      CurBB->splitBasicBlock(CurBBTI, X->getName() + ".atomic.exit");
  BasicBlock *ContBB = CurBB->splitBasicBlock(CurBB->getTerminator(),
                                              X->getName() + ".atomic.cont");
  ContBB->getTerminator()->eraseFromParent();
  Builder.restoreIP(AllocaIP);
  AllocaInst *NewAtomicAddr = Builder.CreateAlloca(XElemTy);
  NewAtomicAddr->setName(X->getName() + "x.new.val");
  Builder.SetInsertPoint(ContBB);
  llvm::PHINode *PHI = Builder.CreatePHI(OldVal->getType(), 2);
  PHI->addIncoming(OldVal, CurBB);
  IntegerType *NewAtomicCastTy =
      IntegerType::get(M.getContext(), XElemTy->getScalarSizeInBits());
  bool IsIntTy = XElemTy->isIntegerTy();
  Value *NewAtomicIntAddr =
      (IsIntTy)
          ? NewAtomicAddr
          : Builder.CreateBitCast(NewAtomicAddr,
                                  NewAtomicCastTy->getPointerTo(Addrspace));
  Value *OldExprVal = PHI;
  if (!IsIntTy) {
    if (XElemTy->isFloatingPointTy()) {
      OldExprVal = Builder.CreateBitCast(PHI, XElemTy,
                                         X->getName() + ".atomic.fltCast");
    } else {
      OldExprVal = Builder.CreateIntToPtr(PHI, XElemTy,
                                          X->getName() + ".atomic.ptrCast");
    }
  }

  Value *Upd = UpdateOp(OldExprVal, Builder);
  Builder.CreateStore(Upd, NewAtomicAddr);
  LoadInst *DesiredVal = Builder.CreateLoad(IntCastTy, NewAtomicIntAddr);
  Value *XAddr =
      (IsIntTy)
          ? X
          : Builder.CreateBitCast(X, IntCastTy->getPointerTo(Addrspace));
  AtomicOrdering Failure =
      llvm::AtomicCmpXchgInst::getStrongestFailureOrdering(AO);
  AtomicCmpXchgInst *Result = Builder.CreateAtomicCmpXchg(
      XAddr, PHI, DesiredVal, llvm::MaybeAlign(), AO, Failure);
  Result->setVolatile(VolatileX);
  Value *PreviousVal = Builder.CreateExtractValue(Result, /*Idxs=*/0);
  Value *SuccessFailureVal = Builder.CreateExtractValue(Result, /*Idxs=*/1);
  PHI->addIncoming(PreviousVal, Builder.GetInsertBlock());
  Builder.CreateCondBr(SuccessFailureVal, ExitBB, ContBB);

  Res.first = OldExprVal;
  Res.second = Upd;

  // set Insertion point in exit block
  if (UnreachableInst *ExitTI =
          dyn_cast<UnreachableInst>(ExitBB->getTerminator())) {
    CurBBTI->eraseFromParent();
    Builder.SetInsertPoint(ExitBB);
  } else {
    Builder.SetInsertPoint(ExitTI);
  }
}

return Res;
4680}

4682OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::createAtomicCapture(
  const LocationDescription &Loc, InsertPointTy AllocaIP, AtomicOpValue &X,
  AtomicOpValue &V, Value *Expr, AtomicOrdering AO,
  AtomicRMWInst::BinOp RMWOp, AtomicUpdateCallbackTy &UpdateOp,
  bool UpdateExpr, bool IsPostfixUpdate, bool IsXBinopExpr) {
if (!updateToLocation(Loc))
  return Loc.IP;

LLVM_DEBUG({do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4693, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic capture expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic capture expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4697, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4699, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
  Type *XTy = X.Var->getType();do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4693, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic capture expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic capture expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4697, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4699, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
  assert(XTy->isPointerTy() &&do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4693, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic capture expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic capture expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4697, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4699, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
         "OMP Atomic expects a pointer to target memory");do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4693, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic capture expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic capture expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4697, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4699, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
  Type *XElemTy = X.ElemTy;do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4693, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic capture expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic capture expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4697, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4699, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
  assert((XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() ||do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4693, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic capture expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic capture expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4697, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4699, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
          XElemTy->isPointerTy()) &&do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4693, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic capture expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic capture expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4697, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4699, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
         "OMP atomic capture expected a scalar type");do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4693, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic capture expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic capture expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4697, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4699, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
  assert((RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) &&do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4693, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic capture expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic capture expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4697, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4699, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
         "OpenMP atomic does not support LT or GT operations");do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4693, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic capture expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic capture expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4697, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4699, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false)
})do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType
("openmp-ir-builder")) { { Type *XTy = X.Var->getType(); (
static_cast <bool> (XTy->isPointerTy() && "OMP Atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("XTy->isPointerTy() && \"OMP Atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4693, __extension__
 __PRETTY_FUNCTION__)); Type *XElemTy = X.ElemTy; (static_cast
 <bool> ((XElemTy->isFloatingPointTy() || XElemTy->
isIntegerTy() || XElemTy->isPointerTy()) && "OMP atomic capture expected a scalar type"
) ? void (0) : __assert_fail ("(XElemTy->isFloatingPointTy() || XElemTy->isIntegerTy() || XElemTy->isPointerTy()) && \"OMP atomic capture expected a scalar type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4697, __extension__
 __PRETTY_FUNCTION__)); (static_cast <bool> ((RMWOp != AtomicRMWInst
::Max) && (RMWOp != AtomicRMWInst::Min) && "OpenMP atomic does not support LT or GT operations"
) ? void (0) : __assert_fail ("(RMWOp != AtomicRMWInst::Max) && (RMWOp != AtomicRMWInst::Min) && \"OpenMP atomic does not support LT or GT operations\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4699, __extension__
 __PRETTY_FUNCTION__)); }; } } while (false);

// If UpdateExpr is 'x' updated with some `expr` not based on 'x',
// 'x' is simply atomically rewritten with 'expr'.
AtomicRMWInst::BinOp AtomicOp = (UpdateExpr ? RMWOp : AtomicRMWInst::Xchg);
std::pair<Value *, Value *> Result =
    emitAtomicUpdate(AllocaIP, X.Var, X.ElemTy, Expr, AO, AtomicOp, UpdateOp,
                     X.IsVolatile, IsXBinopExpr);

Value *CapturedVal = (IsPostfixUpdate ? Result.first : Result.second);
Builder.CreateStore(CapturedVal, V.Var, V.IsVolatile);

checkAndEmitFlushAfterAtomic(Loc, AO, AtomicKind::Capture);
return Builder.saveIP();
4714}

4716OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::createAtomicCompare(
  const LocationDescription &Loc, AtomicOpValue &X, AtomicOpValue &V,
  AtomicOpValue &R, Value *E, Value *D, AtomicOrdering AO,
  omp::OMPAtomicCompareOp Op, bool IsXBinopExpr, bool IsPostfixUpdate,
  bool IsFailOnly) {

if (!updateToLocation(Loc))
  return Loc.IP;

assert(X.Var->getType()->isPointerTy() &&(static_cast <bool> (X.Var->getType()->isPointerTy
() && "OMP atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("X.Var->getType()->isPointerTy() && \"OMP atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4726, __extension__
 __PRETTY_FUNCTION__))
       "OMP atomic expects a pointer to target memory")(static_cast <bool> (X.Var->getType()->isPointerTy
() && "OMP atomic expects a pointer to target memory"
) ? void (0) : __assert_fail ("X.Var->getType()->isPointerTy() && \"OMP atomic expects a pointer to target memory\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4726, __extension__
 __PRETTY_FUNCTION__));
// compare capture
if (V.Var) {
  assert(V.Var->getType()->isPointerTy() && "v.var must be of pointer type")(static_cast <bool> (V.Var->getType()->isPointerTy
() && "v.var must be of pointer type") ? void (0) : __assert_fail
 ("V.Var->getType()->isPointerTy() && \"v.var must be of pointer type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4729, __extension__
 __PRETTY_FUNCTION__));
  assert(V.ElemTy == X.ElemTy && "x and v must be of same type")(static_cast <bool> (V.ElemTy == X.ElemTy && "x and v must be of same type"
) ? void (0) : __assert_fail ("V.ElemTy == X.ElemTy && \"x and v must be of same type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4730, __extension__
 __PRETTY_FUNCTION__));
}

bool IsInteger = E->getType()->isIntegerTy();

if (Op == OMPAtomicCompareOp::EQ) {
  AtomicOrdering Failure = AtomicCmpXchgInst::getStrongestFailureOrdering(AO);
  AtomicCmpXchgInst *Result = nullptr;
  if (!IsInteger) {
    unsigned Addrspace =
        cast<PointerType>(X.Var->getType())->getAddressSpace();
    IntegerType *IntCastTy =
        IntegerType::get(M.getContext(), X.ElemTy->getScalarSizeInBits());
    Value *XBCast =
        Builder.CreateBitCast(X.Var, IntCastTy->getPointerTo(Addrspace));
    Value *EBCast = Builder.CreateBitCast(E, IntCastTy);
    Value *DBCast = Builder.CreateBitCast(D, IntCastTy);
    Result = Builder.CreateAtomicCmpXchg(XBCast, EBCast, DBCast, MaybeAlign(),
                                         AO, Failure);
  } else {
    Result =
        Builder.CreateAtomicCmpXchg(X.Var, E, D, MaybeAlign(), AO, Failure);
  }

  if (V.Var) {
    Value *OldValue = Builder.CreateExtractValue(Result, /*Idxs=*/0);
    if (!IsInteger)
      OldValue = Builder.CreateBitCast(OldValue, X.ElemTy);
    assert(OldValue->getType() == V.ElemTy &&(static_cast <bool> (OldValue->getType() == V.ElemTy
 && "OldValue and V must be of same type") ? void (0)
 : __assert_fail ("OldValue->getType() == V.ElemTy && \"OldValue and V must be of same type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4759, __extension__
 __PRETTY_FUNCTION__))
           "OldValue and V must be of same type")(static_cast <bool> (OldValue->getType() == V.ElemTy
 && "OldValue and V must be of same type") ? void (0)
 : __assert_fail ("OldValue->getType() == V.ElemTy && \"OldValue and V must be of same type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4759, __extension__
 __PRETTY_FUNCTION__));
    if (IsPostfixUpdate) {
      Builder.CreateStore(OldValue, V.Var, V.IsVolatile);
    } else {
      Value *SuccessOrFail = Builder.CreateExtractValue(Result, /*Idxs=*/1);
      if (IsFailOnly) {
        // CurBB----
        //   |     |
        //   v     |
        // ContBB  |
        //   |     |
        //   v     |
        // ExitBB <-
        //
        // where ContBB only contains the store of old value to 'v'.
        BasicBlock *CurBB = Builder.GetInsertBlock();
        Instruction *CurBBTI = CurBB->getTerminator();
        CurBBTI = CurBBTI ? CurBBTI : Builder.CreateUnreachable();
        BasicBlock *ExitBB = CurBB->splitBasicBlock(
            CurBBTI, X.Var->getName() + ".atomic.exit");
        BasicBlock *ContBB = CurBB->splitBasicBlock(
            CurBB->getTerminator(), X.Var->getName() + ".atomic.cont");
        ContBB->getTerminator()->eraseFromParent();
        CurBB->getTerminator()->eraseFromParent();

        Builder.CreateCondBr(SuccessOrFail, ExitBB, ContBB);

        Builder.SetInsertPoint(ContBB);
        Builder.CreateStore(OldValue, V.Var);
        Builder.CreateBr(ExitBB);

        if (UnreachableInst *ExitTI =
                dyn_cast<UnreachableInst>(ExitBB->getTerminator())) {
          CurBBTI->eraseFromParent();
          Builder.SetInsertPoint(ExitBB);
        } else {
          Builder.SetInsertPoint(ExitTI);
        }
      } else {
        Value *CapturedValue =
            Builder.CreateSelect(SuccessOrFail, E, OldValue);
        Builder.CreateStore(CapturedValue, V.Var, V.IsVolatile);
      }
    }
  }
  // The comparison result has to be stored.
  if (R.Var) {
    assert(R.Var->getType()->isPointerTy() &&(static_cast <bool> (R.Var->getType()->isPointerTy
() && "r.var must be of pointer type") ? void (0) : __assert_fail
 ("R.Var->getType()->isPointerTy() && \"r.var must be of pointer type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4807, __extension__
 __PRETTY_FUNCTION__))
           "r.var must be of pointer type")(static_cast <bool> (R.Var->getType()->isPointerTy
() && "r.var must be of pointer type") ? void (0) : __assert_fail
 ("R.Var->getType()->isPointerTy() && \"r.var must be of pointer type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4807, __extension__
 __PRETTY_FUNCTION__));
    assert(R.ElemTy->isIntegerTy() && "r must be of integral type")(static_cast <bool> (R.ElemTy->isIntegerTy() &&
 "r must be of integral type") ? void (0) : __assert_fail ("R.ElemTy->isIntegerTy() && \"r must be of integral type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4808, __extension__
 __PRETTY_FUNCTION__));

    Value *SuccessFailureVal = Builder.CreateExtractValue(Result, /*Idxs=*/1);
    Value *ResultCast = R.IsSigned
                            ? Builder.CreateSExt(SuccessFailureVal, R.ElemTy)
                            : Builder.CreateZExt(SuccessFailureVal, R.ElemTy);
    Builder.CreateStore(ResultCast, R.Var, R.IsVolatile);
  }
} else {
  assert((Op == OMPAtomicCompareOp::MAX || Op == OMPAtomicCompareOp::MIN) &&(static_cast <bool> ((Op == OMPAtomicCompareOp::MAX || Op
 == OMPAtomicCompareOp::MIN) && "Op should be either max or min at this point"
) ? void (0) : __assert_fail ("(Op == OMPAtomicCompareOp::MAX || Op == OMPAtomicCompareOp::MIN) && \"Op should be either max or min at this point\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4818, __extension__
 __PRETTY_FUNCTION__))
         "Op should be either max or min at this point")(static_cast <bool> ((Op == OMPAtomicCompareOp::MAX || Op
 == OMPAtomicCompareOp::MIN) && "Op should be either max or min at this point"
) ? void (0) : __assert_fail ("(Op == OMPAtomicCompareOp::MAX || Op == OMPAtomicCompareOp::MIN) && \"Op should be either max or min at this point\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4818, __extension__
 __PRETTY_FUNCTION__));
  assert(!IsFailOnly && "IsFailOnly is only valid when the comparison is ==")(static_cast <bool> (!IsFailOnly && "IsFailOnly is only valid when the comparison is =="
) ? void (0) : __assert_fail ("!IsFailOnly && \"IsFailOnly is only valid when the comparison is ==\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4819, __extension__
 __PRETTY_FUNCTION__));

  // Reverse the ordop as the OpenMP forms are different from LLVM forms.
  // Let's take max as example.
  // OpenMP form:
  // x = x > expr ? expr : x;
  // LLVM form:
  // *ptr = *ptr > val ? *ptr : val;
  // We need to transform to LLVM form.
  // x = x <= expr ? x : expr;
  AtomicRMWInst::BinOp NewOp;
  if (IsXBinopExpr) {
    if (IsInteger) {
      if (X.IsSigned)
        NewOp = Op == OMPAtomicCompareOp::MAX ? AtomicRMWInst::Min
                                              : AtomicRMWInst::Max;
      else
        NewOp = Op == OMPAtomicCompareOp::MAX ? AtomicRMWInst::UMin
                                              : AtomicRMWInst::UMax;
    } else {
      NewOp = Op == OMPAtomicCompareOp::MAX ? AtomicRMWInst::FMin
                                            : AtomicRMWInst::FMax;
    }
  } else {
    if (IsInteger) {
      if (X.IsSigned)
        NewOp = Op == OMPAtomicCompareOp::MAX ? AtomicRMWInst::Max
                                              : AtomicRMWInst::Min;
      else
        NewOp = Op == OMPAtomicCompareOp::MAX ? AtomicRMWInst::UMax
                                              : AtomicRMWInst::UMin;
    } else {
      NewOp = Op == OMPAtomicCompareOp::MAX ? AtomicRMWInst::FMax
                                            : AtomicRMWInst::FMin;
    }
  }

  AtomicRMWInst *OldValue =
      Builder.CreateAtomicRMW(NewOp, X.Var, E, MaybeAlign(), AO);
  if (V.Var) {
    Value *CapturedValue = nullptr;
    if (IsPostfixUpdate) {
      CapturedValue = OldValue;
    } else {
      CmpInst::Predicate Pred;
      switch (NewOp) {
      case AtomicRMWInst::Max:
        Pred = CmpInst::ICMP_SGT;
        break;
      case AtomicRMWInst::UMax:
        Pred = CmpInst::ICMP_UGT;
        break;
      case AtomicRMWInst::FMax:
        Pred = CmpInst::FCMP_OGT;
        break;
      case AtomicRMWInst::Min:
        Pred = CmpInst::ICMP_SLT;
        break;
      case AtomicRMWInst::UMin:
        Pred = CmpInst::ICMP_ULT;
        break;
      case AtomicRMWInst::FMin:
        Pred = CmpInst::FCMP_OLT;
        break;
      default:
        llvm_unreachable("unexpected comparison op")::llvm::llvm_unreachable_internal("unexpected comparison op",
 "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 4884);
      }
      Value *NonAtomicCmp = Builder.CreateCmp(Pred, OldValue, E);
      CapturedValue = Builder.CreateSelect(NonAtomicCmp, E, OldValue);
    }
    Builder.CreateStore(CapturedValue, V.Var, V.IsVolatile);
  }
}

checkAndEmitFlushAfterAtomic(Loc, AO, AtomicKind::Compare);

return Builder.saveIP();
4896}

4898GlobalVariable *
4899OpenMPIRBuilder::createOffloadMapnames(SmallVectorImpl<llvm::Constant *> &Names,
                                     std::string VarName) {
llvm::Constant *MapNamesArrayInit = llvm::ConstantArray::get(
    llvm::ArrayType::get(
        llvm::Type::getInt8Ty(M.getContext())->getPointerTo(), Names.size()),
    Names);
auto *MapNamesArrayGlobal = new llvm::GlobalVariable(
    M, MapNamesArrayInit->getType(),
    /*isConstant=*/true, llvm::GlobalValue::PrivateLinkage, MapNamesArrayInit,
    VarName);
return MapNamesArrayGlobal;
4910}

4912// Create all simple and struct types exposed by the runtime and remember
4913// the llvm::PointerTypes of them for easy access later.
4914void OpenMPIRBuilder::initializeTypes(Module &M) {
LLVMContext &Ctx = M.getContext();
StructType *T;
4917#define OMP_TYPE(VarName, InitValue) VarName = InitValue;
4918#define OMP_ARRAY_TYPE(VarName, ElemTy, ArraySize)                             \
VarName##Ty = ArrayType::get(ElemTy, ArraySize);                             \
VarName##PtrTy = PointerType::getUnqual(VarName##Ty);
4921#define OMP_FUNCTION_TYPE(VarName, IsVarArg, ReturnType, ...)                  \
VarName = FunctionType::get(ReturnType, {__VA_ARGS__}, IsVarArg);            \
VarName##Ptr = PointerType::getUnqual(VarName);
4924#define OMP_STRUCT_TYPE(VarName, StructName, Packed, ...)                      \
T = StructType::getTypeByName(Ctx, StructName);                              \
if (!T)                                                                      \
  T = StructType::create(Ctx, {__VA_ARGS__}, StructName, Packed);            \
VarName = T;                                                                 \
VarName##Ptr = PointerType::getUnqual(T);
4930#include "llvm/Frontend/OpenMP/OMPKinds.def"
4931}

4933void OpenMPIRBuilder::OutlineInfo::collectBlocks(
  SmallPtrSetImpl<BasicBlock *> &BlockSet,
  SmallVectorImpl<BasicBlock *> &BlockVector) {
SmallVector<BasicBlock *, 32> Worklist;
BlockSet.insert(EntryBB);
BlockSet.insert(ExitBB);

Worklist.push_back(EntryBB);
while (!Worklist.empty()) {
  BasicBlock *BB = Worklist.pop_back_val();
  BlockVector.push_back(BB);
  for (BasicBlock *SuccBB : successors(BB))
    if (BlockSet.insert(SuccBB).second)
      Worklist.push_back(SuccBB);
}
4948}

4950void OpenMPIRBuilder::createOffloadEntry(Constant *ID, Constant *Addr,
                                       uint64_t Size, int32_t Flags,
                                       GlobalValue::LinkageTypes) {
if (!Config.isTargetCodegen()) {
  emitOffloadingEntry(ID, Addr->getName(), Size, Flags);
  return;
}
// TODO: Add support for global variables on the device after declare target
// support.
Function *Fn = dyn_cast<Function>(Addr);
if (!Fn)
  return;

Module &M = *(Fn->getParent());
LLVMContext &Ctx = M.getContext();

// Get "nvvm.annotations" metadata node.
NamedMDNode *MD = M.getOrInsertNamedMetadata("nvvm.annotations");

Metadata *MDVals[] = {
    ConstantAsMetadata::get(Fn), MDString::get(Ctx, "kernel"),
    ConstantAsMetadata::get(ConstantInt::get(Type::getInt32Ty(Ctx), 1))};
// Append metadata to nvvm.annotations.
MD->addOperand(MDNode::get(Ctx, MDVals));

// Add a function attribute for the kernel.
Fn->addFnAttr(Attribute::get(Ctx, "kernel"));
if (Triple(M.getTargetTriple()).isAMDGCN())
  Fn->addFnAttr("uniform-work-group-size", "true");
4979}

4981// We only generate metadata for function that contain target regions.
4982void OpenMPIRBuilder::createOffloadEntriesAndInfoMetadata(
  EmitMetadataErrorReportFunctionTy &ErrorFn) {

// If there are no entries, we don't need to do anything.
if (OffloadInfoManager.empty())
  return;

LLVMContext &C = M.getContext();
SmallVector<std::pair<const OffloadEntriesInfoManager::OffloadEntryInfo *,
                      TargetRegionEntryInfo>,
            16>
    OrderedEntries(OffloadInfoManager.size());

// Auxiliary methods to create metadata values and strings.
auto &&GetMDInt = [this](unsigned V) {
  return ConstantAsMetadata::get(ConstantInt::get(Builder.getInt32Ty(), V));
};

auto &&GetMDString = [&C](StringRef V) { return MDString::get(C, V); };

// Create the offloading info metadata node.
NamedMDNode *MD = M.getOrInsertNamedMetadata("omp_offload.info");
auto &&TargetRegionMetadataEmitter =
    [&C, MD, &OrderedEntries, &GetMDInt, &GetMDString](
        const TargetRegionEntryInfo &EntryInfo,
        const OffloadEntriesInfoManager::OffloadEntryInfoTargetRegion &E) {
      // Generate metadata for target regions. Each entry of this metadata
      // contains:
      // - Entry 0 -> Kind of this type of metadata (0).
      // - Entry 1 -> Device ID of the file where the entry was identified.
      // - Entry 2 -> File ID of the file where the entry was identified.
      // - Entry 3 -> Mangled name of the function where the entry was
      // identified.
      // - Entry 4 -> Line in the file where the entry was identified.
      // - Entry 5 -> Count of regions at this DeviceID/FilesID/Line.
      // - Entry 6 -> Order the entry was created.
      // The first element of the metadata node is the kind.
      Metadata *Ops[] = {
          GetMDInt(E.getKind()),      GetMDInt(EntryInfo.DeviceID),
          GetMDInt(EntryInfo.FileID), GetMDString(EntryInfo.ParentName),
          GetMDInt(EntryInfo.Line),   GetMDInt(EntryInfo.Count),
          GetMDInt(E.getOrder())};

      // Save this entry in the right position of the ordered entries array.
      OrderedEntries[E.getOrder()] = std::make_pair(&E, EntryInfo);

      // Add metadata to the named metadata node.
      MD->addOperand(MDNode::get(C, Ops));
    };

OffloadInfoManager.actOnTargetRegionEntriesInfo(TargetRegionMetadataEmitter);

// Create function that emits metadata for each device global variable entry;
auto &&DeviceGlobalVarMetadataEmitter =
    [&C, &OrderedEntries, &GetMDInt, &GetMDString, MD](
        StringRef MangledName,
        const OffloadEntriesInfoManager::OffloadEntryInfoDeviceGlobalVar &E) {
      // Generate metadata for global variables. Each entry of this metadata
      // contains:
      // - Entry 0 -> Kind of this type of metadata (1).
      // - Entry 1 -> Mangled name of the variable.
      // - Entry 2 -> Declare target kind.
      // - Entry 3 -> Order the entry was created.
      // The first element of the metadata node is the kind.
      Metadata *Ops[] = {GetMDInt(E.getKind()), GetMDString(MangledName),
                         GetMDInt(E.getFlags()), GetMDInt(E.getOrder())};

      // Save this entry in the right position of the ordered entries array.
      TargetRegionEntryInfo varInfo(MangledName, 0, 0, 0);
      OrderedEntries[E.getOrder()] = std::make_pair(&E, varInfo);

      // Add metadata to the named metadata node.
      MD->addOperand(MDNode::get(C, Ops));
    };

OffloadInfoManager.actOnDeviceGlobalVarEntriesInfo(
    DeviceGlobalVarMetadataEmitter);

for (const auto &E : OrderedEntries) {
  assert(E.first && "All ordered entries must exist!")(static_cast <bool> (E.first && "All ordered entries must exist!"
) ? void (0) : __assert_fail ("E.first && \"All ordered entries must exist!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5061, __extension__
 __PRETTY_FUNCTION__));
  if (const auto *CE =
          dyn_cast<OffloadEntriesInfoManager::OffloadEntryInfoTargetRegion>(
              E.first)) {
    if (!CE->getID() || !CE->getAddress()) {
      // Do not blame the entry if the parent funtion is not emitted.
      TargetRegionEntryInfo EntryInfo = E.second;
      StringRef FnName = EntryInfo.ParentName;
      if (!M.getNamedValue(FnName))
        continue;
      ErrorFn(EMIT_MD_TARGET_REGION_ERROR, EntryInfo);
      continue;
    }
    createOffloadEntry(CE->getID(), CE->getAddress(),
                       /*Size=*/0, CE->getFlags(),
                       GlobalValue::WeakAnyLinkage);
  } else if (const auto *CE = dyn_cast<
                 OffloadEntriesInfoManager::OffloadEntryInfoDeviceGlobalVar>(
                 E.first)) {
    OffloadEntriesInfoManager::OMPTargetGlobalVarEntryKind Flags =
        static_cast<OffloadEntriesInfoManager::OMPTargetGlobalVarEntryKind>(
            CE->getFlags());
    switch (Flags) {
    case OffloadEntriesInfoManager::OMPTargetGlobalVarEntryTo: {
      if (Config.isEmbedded() && Config.hasRequiresUnifiedSharedMemory())
        continue;
      if (!CE->getAddress()) {
        ErrorFn(EMIT_MD_DECLARE_TARGET_ERROR, E.second);
        continue;
      }
      // The vaiable has no definition - no need to add the entry.
      if (CE->getVarSize() == 0)
        continue;
      break;
    }
    case OffloadEntriesInfoManager::OMPTargetGlobalVarEntryLink:
      assert(((Config.isEmbedded() && !CE->getAddress()) ||(static_cast <bool> (((Config.isEmbedded() && !
CE->getAddress()) || (!Config.isEmbedded() && CE->
getAddress())) && "Declaret target link address is set."
) ? void (0) : __assert_fail ("((Config.isEmbedded() && !CE->getAddress()) || (!Config.isEmbedded() && CE->getAddress())) && \"Declaret target link address is set.\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5099, __extension__
 __PRETTY_FUNCTION__))
              (!Config.isEmbedded() && CE->getAddress())) &&(static_cast <bool> (((Config.isEmbedded() && !
CE->getAddress()) || (!Config.isEmbedded() && CE->
getAddress())) && "Declaret target link address is set."
) ? void (0) : __assert_fail ("((Config.isEmbedded() && !CE->getAddress()) || (!Config.isEmbedded() && CE->getAddress())) && \"Declaret target link address is set.\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5099, __extension__
 __PRETTY_FUNCTION__))
             "Declaret target link address is set.")(static_cast <bool> (((Config.isEmbedded() && !
CE->getAddress()) || (!Config.isEmbedded() && CE->
getAddress())) && "Declaret target link address is set."
) ? void (0) : __assert_fail ("((Config.isEmbedded() && !CE->getAddress()) || (!Config.isEmbedded() && CE->getAddress())) && \"Declaret target link address is set.\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5099, __extension__
 __PRETTY_FUNCTION__));
      if (Config.isEmbedded())
        continue;
      if (!CE->getAddress()) {
        ErrorFn(EMIT_MD_GLOBAL_VAR_LINK_ERROR, TargetRegionEntryInfo());
        continue;
      }
      break;
    }

    // Hidden or internal symbols on the device are not externally visible.
    // We should not attempt to register them by creating an offloading
    // entry.
    if (auto *GV = dyn_cast<GlobalValue>(CE->getAddress()))
      if (GV->hasLocalLinkage() || GV->hasHiddenVisibility())
        continue;

    createOffloadEntry(CE->getAddress(), CE->getAddress(), CE->getVarSize(),
                       Flags, CE->getLinkage());

  } else {
    llvm_unreachable("Unsupported entry kind.")::llvm::llvm_unreachable_internal("Unsupported entry kind.", "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp"
, 5120);
  }
}
5123}

5125void TargetRegionEntryInfo::getTargetRegionEntryFnName(
  SmallVectorImpl<char> &Name, StringRef ParentName, unsigned DeviceID,
  unsigned FileID, unsigned Line, unsigned Count) {
raw_svector_ostream OS(Name);
OS << "__omp_offloading" << llvm::format("_%x", DeviceID)
   << llvm::format("_%x_", FileID) << ParentName << "_l" << Line;
if (Count)
  OS << "_" << Count;
5133}

5135void OffloadEntriesInfoManager::getTargetRegionEntryFnName(
  SmallVectorImpl<char> &Name, const TargetRegionEntryInfo &EntryInfo) {
unsigned NewCount = getTargetRegionEntryInfoCount(EntryInfo);
TargetRegionEntryInfo::getTargetRegionEntryFnName(
    Name, EntryInfo.ParentName, EntryInfo.DeviceID, EntryInfo.FileID,
    EntryInfo.Line, NewCount);
5141}

5143/// Loads all the offload entries information from the host IR
5144/// metadata.
5145void OpenMPIRBuilder::loadOffloadInfoMetadata(Module &M) {
// If we are in target mode, load the metadata from the host IR. This code has
// to match the metadata creation in createOffloadEntriesAndInfoMetadata().

NamedMDNode *MD = M.getNamedMetadata(ompOffloadInfoName);
if (!MD)
  return;

for (MDNode *MN : MD->operands()) {
  auto &&GetMDInt = [MN](unsigned Idx) {
    auto *V = cast<ConstantAsMetadata>(MN->getOperand(Idx));
    return cast<ConstantInt>(V->getValue())->getZExtValue();
  };

  auto &&GetMDString = [MN](unsigned Idx) {
    auto *V = cast<MDString>(MN->getOperand(Idx));
    return V->getString();
  };

  switch (GetMDInt(0)) {
  default:
    llvm_unreachable("Unexpected metadata!")::llvm::llvm_unreachable_internal("Unexpected metadata!", "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp"
, 5166);
    break;
  case OffloadEntriesInfoManager::OffloadEntryInfo::
      OffloadingEntryInfoTargetRegion: {
    TargetRegionEntryInfo EntryInfo(/*ParentName=*/GetMDString(3),
                                    /*DeviceID=*/GetMDInt(1),
                                    /*FileID=*/GetMDInt(2),
                                    /*Line=*/GetMDInt(4),
                                    /*Count=*/GetMDInt(5));
    OffloadInfoManager.initializeTargetRegionEntryInfo(EntryInfo,
                                                       /*Order=*/GetMDInt(6));
    break;
  }
  case OffloadEntriesInfoManager::OffloadEntryInfo::
      OffloadingEntryInfoDeviceGlobalVar:
    OffloadInfoManager.initializeDeviceGlobalVarEntryInfo(
        /*MangledName=*/GetMDString(1),
        static_cast<OffloadEntriesInfoManager::OMPTargetGlobalVarEntryKind>(
            /*Flags=*/GetMDInt(2)),
        /*Order=*/GetMDInt(3));
    break;
  }
}
5189}

5191bool OffloadEntriesInfoManager::empty() const {
return OffloadEntriesTargetRegion.empty() &&
       OffloadEntriesDeviceGlobalVar.empty();
5194}

5196unsigned OffloadEntriesInfoManager::getTargetRegionEntryInfoCount(
  const TargetRegionEntryInfo &EntryInfo) const {
auto It = OffloadEntriesTargetRegionCount.find(
    getTargetRegionEntryCountKey(EntryInfo));
if (It == OffloadEntriesTargetRegionCount.end())
  return 0;
return It->second;
5203}

5205void OffloadEntriesInfoManager::incrementTargetRegionEntryInfoCount(
  const TargetRegionEntryInfo &EntryInfo) {
OffloadEntriesTargetRegionCount[getTargetRegionEntryCountKey(EntryInfo)] =
    EntryInfo.Count + 1;
5209}

5211/// Initialize target region entry.
5212void OffloadEntriesInfoManager::initializeTargetRegionEntryInfo(
  const TargetRegionEntryInfo &EntryInfo, unsigned Order) {
OffloadEntriesTargetRegion[EntryInfo] =
    OffloadEntryInfoTargetRegion(Order, /*Addr=*/nullptr, /*ID=*/nullptr,
                                 OMPTargetRegionEntryTargetRegion);
++OffloadingEntriesNum;
5218}

5220void OffloadEntriesInfoManager::registerTargetRegionEntryInfo(
  TargetRegionEntryInfo EntryInfo, Constant *Addr, Constant *ID,
  OMPTargetRegionEntryKind Flags) {
assert(EntryInfo.Count == 0 && "expected default EntryInfo")(static_cast <bool> (EntryInfo.Count == 0 && "expected default EntryInfo"
) ? void (0) : __assert_fail ("EntryInfo.Count == 0 && \"expected default EntryInfo\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5223, __extension__
 __PRETTY_FUNCTION__));

// Update the EntryInfo with the next available count for this location.
EntryInfo.Count = getTargetRegionEntryInfoCount(EntryInfo);

// If we are emitting code for a target, the entry is already initialized,
// only has to be registered.
if (OMPBuilder->Config.isEmbedded()) {
  // This could happen if the device compilation is invoked standalone.
  if (!hasTargetRegionEntryInfo(EntryInfo)) {
    return;
  }
  auto &Entry = OffloadEntriesTargetRegion[EntryInfo];
  Entry.setAddress(Addr);
  Entry.setID(ID);
  Entry.setFlags(Flags);
} else {
  if (Flags == OffloadEntriesInfoManager::OMPTargetRegionEntryTargetRegion &&
      hasTargetRegionEntryInfo(EntryInfo, /*IgnoreAddressId*/ true))
    return;
  assert(!hasTargetRegionEntryInfo(EntryInfo) &&(static_cast <bool> (!hasTargetRegionEntryInfo(EntryInfo
) && "Target region entry already registered!") ? void
 (0) : __assert_fail ("!hasTargetRegionEntryInfo(EntryInfo) && \"Target region entry already registered!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5244, __extension__
 __PRETTY_FUNCTION__))
         "Target region entry already registered!")(static_cast <bool> (!hasTargetRegionEntryInfo(EntryInfo
) && "Target region entry already registered!") ? void
 (0) : __assert_fail ("!hasTargetRegionEntryInfo(EntryInfo) && \"Target region entry already registered!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5244, __extension__
 __PRETTY_FUNCTION__));
  OffloadEntryInfoTargetRegion Entry(OffloadingEntriesNum, Addr, ID, Flags);
  OffloadEntriesTargetRegion[EntryInfo] = Entry;
  ++OffloadingEntriesNum;
}
incrementTargetRegionEntryInfoCount(EntryInfo);
5250}

5252bool OffloadEntriesInfoManager::hasTargetRegionEntryInfo(
  TargetRegionEntryInfo EntryInfo, bool IgnoreAddressId) const {

// Update the EntryInfo with the next available count for this location.
EntryInfo.Count = getTargetRegionEntryInfoCount(EntryInfo);

auto It = OffloadEntriesTargetRegion.find(EntryInfo);
if (It == OffloadEntriesTargetRegion.end()) {
  return false;
}
// Fail if this entry is already registered.
if (!IgnoreAddressId && (It->second.getAddress() || It->second.getID()))
  return false;
return true;
5266}

5268void OffloadEntriesInfoManager::actOnTargetRegionEntriesInfo(
  const OffloadTargetRegionEntryInfoActTy &Action) {
// Scan all target region entries and perform the provided action.
for (const auto &It : OffloadEntriesTargetRegion) {
  Action(It.first, It.second);
}
5274}

5276void OffloadEntriesInfoManager::initializeDeviceGlobalVarEntryInfo(
  StringRef Name, OMPTargetGlobalVarEntryKind Flags, unsigned Order) {
OffloadEntriesDeviceGlobalVar.try_emplace(Name, Order, Flags);
++OffloadingEntriesNum;
5280}

5282void OffloadEntriesInfoManager::registerDeviceGlobalVarEntryInfo(
  StringRef VarName, Constant *Addr, int64_t VarSize,
  OMPTargetGlobalVarEntryKind Flags, GlobalValue::LinkageTypes Linkage) {
if (OMPBuilder->Config.isEmbedded()) {
  // This could happen if the device compilation is invoked standalone.
  if (!hasDeviceGlobalVarEntryInfo(VarName))
    return;
  auto &Entry = OffloadEntriesDeviceGlobalVar[VarName];
  if (Entry.getAddress() && hasDeviceGlobalVarEntryInfo(VarName)) {
    if (Entry.getVarSize() == 0) {
      Entry.setVarSize(VarSize);
      Entry.setLinkage(Linkage);
    }
    return;
  }
  Entry.setVarSize(VarSize);
  Entry.setLinkage(Linkage);
  Entry.setAddress(Addr);
} else {
  if (hasDeviceGlobalVarEntryInfo(VarName)) {
    auto &Entry = OffloadEntriesDeviceGlobalVar[VarName];
    assert(Entry.isValid() && Entry.getFlags() == Flags &&(static_cast <bool> (Entry.isValid() && Entry.getFlags
() == Flags && "Entry not initialized!") ? void (0) :
 __assert_fail ("Entry.isValid() && Entry.getFlags() == Flags && \"Entry not initialized!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5304, __extension__
 __PRETTY_FUNCTION__))
           "Entry not initialized!")(static_cast <bool> (Entry.isValid() && Entry.getFlags
() == Flags && "Entry not initialized!") ? void (0) :
 __assert_fail ("Entry.isValid() && Entry.getFlags() == Flags && \"Entry not initialized!\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5304, __extension__
 __PRETTY_FUNCTION__));
    if (Entry.getVarSize() == 0) {
      Entry.setVarSize(VarSize);
      Entry.setLinkage(Linkage);
    }
    return;
  }
  OffloadEntriesDeviceGlobalVar.try_emplace(VarName, OffloadingEntriesNum,
                                            Addr, VarSize, Flags, Linkage);
  ++OffloadingEntriesNum;
}
5315}

5317void OffloadEntriesInfoManager::actOnDeviceGlobalVarEntriesInfo(
  const OffloadDeviceGlobalVarEntryInfoActTy &Action) {
// Scan all target region entries and perform the provided action.
for (const auto &E : OffloadEntriesDeviceGlobalVar)
  Action(E.getKey(), E.getValue());
5322}

5324void CanonicalLoopInfo::collectControlBlocks(
  SmallVectorImpl<BasicBlock *> &BBs) {
// We only count those BBs as control block for which we do not need to
// reverse the CFG, i.e. not the loop body which can contain arbitrary control
// flow. For consistency, this also means we do not add the Body block, which
// is just the entry to the body code.
BBs.reserve(BBs.size() + 6);
BBs.append({getPreheader(), Header, Cond, Latch, Exit, getAfter()});
5332}

5334BasicBlock *CanonicalLoopInfo::getPreheader() const {
assert(isValid() && "Requires a valid canonical loop")(static_cast <bool> (isValid() && "Requires a valid canonical loop"
) ? void (0) : __assert_fail ("isValid() && \"Requires a valid canonical loop\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5335, __extension__
 __PRETTY_FUNCTION__));
for (BasicBlock *Pred : predecessors(Header)) {
  if (Pred != Latch)
    return Pred;
}
llvm_unreachable("Missing preheader")::llvm::llvm_unreachable_internal("Missing preheader", "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp"
, 5340);
5341}

5343void CanonicalLoopInfo::setTripCount(Value *TripCount) {
assert(isValid() && "Requires a valid canonical loop")(static_cast <bool> (isValid() && "Requires a valid canonical loop"
) ? void (0) : __assert_fail ("isValid() && \"Requires a valid canonical loop\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5344, __extension__
 __PRETTY_FUNCTION__));

Instruction *CmpI = &getCond()->front();
assert(isa<CmpInst>(CmpI) && "First inst must compare IV with TripCount")(static_cast <bool> (isa<CmpInst>(CmpI) &&
 "First inst must compare IV with TripCount") ? void (0) : __assert_fail
 ("isa<CmpInst>(CmpI) && \"First inst must compare IV with TripCount\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5347, __extension__
 __PRETTY_FUNCTION__));
CmpI->setOperand(1, TripCount);

5350#ifndef NDEBUG
assertOK();
5352#endif
5353}

5355void CanonicalLoopInfo::mapIndVar(
  llvm::function_ref<Value *(Instruction *)> Updater) {
assert(isValid() && "Requires a valid canonical loop")(static_cast <bool> (isValid() && "Requires a valid canonical loop"
) ? void (0) : __assert_fail ("isValid() && \"Requires a valid canonical loop\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5357, __extension__
 __PRETTY_FUNCTION__));

Instruction *OldIV = getIndVar();

// Record all uses excluding those introduced by the updater. Uses by the
// CanonicalLoopInfo itself to keep track of the number of iterations are
// excluded.
SmallVector<Use *> ReplacableUses;
for (Use &U : OldIV->uses()) {
  auto *User = dyn_cast<Instruction>(U.getUser());
  if (!User)
    continue;
  if (User->getParent() == getCond())
    continue;
  if (User->getParent() == getLatch())
    continue;
  ReplacableUses.push_back(&U);
}

// Run the updater that may introduce new uses
Value *NewIV = Updater(OldIV);

// Replace the old uses with the value returned by the updater.
for (Use *U : ReplacableUses)
  U->set(NewIV);

5383#ifndef NDEBUG
assertOK();
5385#endif
5386}

5388void CanonicalLoopInfo::assertOK() const {
5389#ifndef NDEBUG
// No constraints if this object currently does not describe a loop.
if (!isValid())
  return;

BasicBlock *Preheader = getPreheader();
BasicBlock *Body = getBody();
BasicBlock *After = getAfter();

// Verify standard control-flow we use for OpenMP loops.
assert(Preheader)(static_cast <bool> (Preheader) ? void (0) : __assert_fail
 ("Preheader", "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5399
, __extension__ __PRETTY_FUNCTION__));
assert(isa<BranchInst>(Preheader->getTerminator()) &&(static_cast <bool> (isa<BranchInst>(Preheader->
getTerminator()) && "Preheader must terminate with unconditional branch"
) ? void (0) : __assert_fail ("isa<BranchInst>(Preheader->getTerminator()) && \"Preheader must terminate with unconditional branch\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5401, __extension__
 __PRETTY_FUNCTION__))
       "Preheader must terminate with unconditional branch")(static_cast <bool> (isa<BranchInst>(Preheader->
getTerminator()) && "Preheader must terminate with unconditional branch"
) ? void (0) : __assert_fail ("isa<BranchInst>(Preheader->getTerminator()) && \"Preheader must terminate with unconditional branch\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5401, __extension__
 __PRETTY_FUNCTION__));
assert(Preheader->getSingleSuccessor() == Header &&(static_cast <bool> (Preheader->getSingleSuccessor()
 == Header && "Preheader must jump to header") ? void
 (0) : __assert_fail ("Preheader->getSingleSuccessor() == Header && \"Preheader must jump to header\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5403, __extension__
 __PRETTY_FUNCTION__))
       "Preheader must jump to header")(static_cast <bool> (Preheader->getSingleSuccessor()
 == Header && "Preheader must jump to header") ? void
 (0) : __assert_fail ("Preheader->getSingleSuccessor() == Header && \"Preheader must jump to header\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5403, __extension__
 __PRETTY_FUNCTION__));

assert(Header)(static_cast <bool> (Header) ? void (0) : __assert_fail
 ("Header", "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5405
, __extension__ __PRETTY_FUNCTION__));
assert(isa<BranchInst>(Header->getTerminator()) &&(static_cast <bool> (isa<BranchInst>(Header->getTerminator
()) && "Header must terminate with unconditional branch"
) ? void (0) : __assert_fail ("isa<BranchInst>(Header->getTerminator()) && \"Header must terminate with unconditional branch\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5407, __extension__
 __PRETTY_FUNCTION__))
       "Header must terminate with unconditional branch")(static_cast <bool> (isa<BranchInst>(Header->getTerminator
()) && "Header must terminate with unconditional branch"
) ? void (0) : __assert_fail ("isa<BranchInst>(Header->getTerminator()) && \"Header must terminate with unconditional branch\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5407, __extension__
 __PRETTY_FUNCTION__));
assert(Header->getSingleSuccessor() == Cond &&(static_cast <bool> (Header->getSingleSuccessor() ==
 Cond && "Header must jump to exiting block") ? void (
0) : __assert_fail ("Header->getSingleSuccessor() == Cond && \"Header must jump to exiting block\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5409, __extension__
 __PRETTY_FUNCTION__))
       "Header must jump to exiting block")(static_cast <bool> (Header->getSingleSuccessor() ==
 Cond && "Header must jump to exiting block") ? void (
0) : __assert_fail ("Header->getSingleSuccessor() == Cond && \"Header must jump to exiting block\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5409, __extension__
 __PRETTY_FUNCTION__));

assert(Cond)(static_cast <bool> (Cond) ? void (0) : __assert_fail (
"Cond", "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5411, __extension__
 __PRETTY_FUNCTION__));
assert(Cond->getSinglePredecessor() == Header &&(static_cast <bool> (Cond->getSinglePredecessor() ==
 Header && "Exiting block only reachable from header"
) ? void (0) : __assert_fail ("Cond->getSinglePredecessor() == Header && \"Exiting block only reachable from header\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5413, __extension__
 __PRETTY_FUNCTION__))
       "Exiting block only reachable from header")(static_cast <bool> (Cond->getSinglePredecessor() ==
 Header && "Exiting block only reachable from header"
) ? void (0) : __assert_fail ("Cond->getSinglePredecessor() == Header && \"Exiting block only reachable from header\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5413, __extension__
 __PRETTY_FUNCTION__));

assert(isa<BranchInst>(Cond->getTerminator()) &&(static_cast <bool> (isa<BranchInst>(Cond->getTerminator
()) && "Exiting block must terminate with conditional branch"
) ? void (0) : __assert_fail ("isa<BranchInst>(Cond->getTerminator()) && \"Exiting block must terminate with conditional branch\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5416, __extension__
 __PRETTY_FUNCTION__))
       "Exiting block must terminate with conditional branch")(static_cast <bool> (isa<BranchInst>(Cond->getTerminator
()) && "Exiting block must terminate with conditional branch"
) ? void (0) : __assert_fail ("isa<BranchInst>(Cond->getTerminator()) && \"Exiting block must terminate with conditional branch\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5416, __extension__
 __PRETTY_FUNCTION__));
assert(size(successors(Cond)) == 2 &&(static_cast <bool> (size(successors(Cond)) == 2 &&
 "Exiting block must have two successors") ? void (0) : __assert_fail
 ("size(successors(Cond)) == 2 && \"Exiting block must have two successors\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5418, __extension__
 __PRETTY_FUNCTION__))
       "Exiting block must have two successors")(static_cast <bool> (size(successors(Cond)) == 2 &&
 "Exiting block must have two successors") ? void (0) : __assert_fail
 ("size(successors(Cond)) == 2 && \"Exiting block must have two successors\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5418, __extension__
 __PRETTY_FUNCTION__));
assert(cast<BranchInst>(Cond->getTerminator())->getSuccessor(0) == Body &&(static_cast <bool> (cast<BranchInst>(Cond->getTerminator
())->getSuccessor(0) == Body && "Exiting block's first successor jump to the body"
) ? void (0) : __assert_fail ("cast<BranchInst>(Cond->getTerminator())->getSuccessor(0) == Body && \"Exiting block's first successor jump to the body\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5420, __extension__
 __PRETTY_FUNCTION__))
       "Exiting block's first successor jump to the body")(static_cast <bool> (cast<BranchInst>(Cond->getTerminator
())->getSuccessor(0) == Body && "Exiting block's first successor jump to the body"
) ? void (0) : __assert_fail ("cast<BranchInst>(Cond->getTerminator())->getSuccessor(0) == Body && \"Exiting block's first successor jump to the body\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5420, __extension__
 __PRETTY_FUNCTION__));
assert(cast<BranchInst>(Cond->getTerminator())->getSuccessor(1) == Exit &&(static_cast <bool> (cast<BranchInst>(Cond->getTerminator
())->getSuccessor(1) == Exit && "Exiting block's second successor must exit the loop"
) ? void (0) : __assert_fail ("cast<BranchInst>(Cond->getTerminator())->getSuccessor(1) == Exit && \"Exiting block's second successor must exit the loop\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5422, __extension__
 __PRETTY_FUNCTION__))
       "Exiting block's second successor must exit the loop")(static_cast <bool> (cast<BranchInst>(Cond->getTerminator
())->getSuccessor(1) == Exit && "Exiting block's second successor must exit the loop"
) ? void (0) : __assert_fail ("cast<BranchInst>(Cond->getTerminator())->getSuccessor(1) == Exit && \"Exiting block's second successor must exit the loop\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5422, __extension__
 __PRETTY_FUNCTION__));

assert(Body)(static_cast <bool> (Body) ? void (0) : __assert_fail (
"Body", "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5424, __extension__
 __PRETTY_FUNCTION__));
assert(Body->getSinglePredecessor() == Cond &&(static_cast <bool> (Body->getSinglePredecessor() ==
 Cond && "Body only reachable from exiting block") ? void
 (0) : __assert_fail ("Body->getSinglePredecessor() == Cond && \"Body only reachable from exiting block\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5426, __extension__
 __PRETTY_FUNCTION__))
       "Body only reachable from exiting block")(static_cast <bool> (Body->getSinglePredecessor() ==
 Cond && "Body only reachable from exiting block") ? void
 (0) : __assert_fail ("Body->getSinglePredecessor() == Cond && \"Body only reachable from exiting block\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5426, __extension__
 __PRETTY_FUNCTION__));
assert(!isa<PHINode>(Body->front()))(static_cast <bool> (!isa<PHINode>(Body->front
())) ? void (0) : __assert_fail ("!isa<PHINode>(Body->front())"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5427, __extension__
 __PRETTY_FUNCTION__));

assert(Latch)(static_cast <bool> (Latch) ? void (0) : __assert_fail (
"Latch", "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5429, __extension__
 __PRETTY_FUNCTION__));
assert(isa<BranchInst>(Latch->getTerminator()) &&(static_cast <bool> (isa<BranchInst>(Latch->getTerminator
()) && "Latch must terminate with unconditional branch"
) ? void (0) : __assert_fail ("isa<BranchInst>(Latch->getTerminator()) && \"Latch must terminate with unconditional branch\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5431, __extension__
 __PRETTY_FUNCTION__))
       "Latch must terminate with unconditional branch")(static_cast <bool> (isa<BranchInst>(Latch->getTerminator
()) && "Latch must terminate with unconditional branch"
) ? void (0) : __assert_fail ("isa<BranchInst>(Latch->getTerminator()) && \"Latch must terminate with unconditional branch\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5431, __extension__
 __PRETTY_FUNCTION__));
assert(Latch->getSingleSuccessor() == Header && "Latch must jump to header")(static_cast <bool> (Latch->getSingleSuccessor() == Header
 && "Latch must jump to header") ? void (0) : __assert_fail
 ("Latch->getSingleSuccessor() == Header && \"Latch must jump to header\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5432, __extension__
 __PRETTY_FUNCTION__));
// TODO: To support simple redirecting of the end of the body code that has
// multiple; introduce another auxiliary basic block like preheader and after.
assert(Latch->getSinglePredecessor() != nullptr)(static_cast <bool> (Latch->getSinglePredecessor() !=
 nullptr) ? void (0) : __assert_fail ("Latch->getSinglePredecessor() != nullptr"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5435, __extension__
 __PRETTY_FUNCTION__));
assert(!isa<PHINode>(Latch->front()))(static_cast <bool> (!isa<PHINode>(Latch->front
())) ? void (0) : __assert_fail ("!isa<PHINode>(Latch->front())"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5436, __extension__
 __PRETTY_FUNCTION__));

assert(Exit)(static_cast <bool> (Exit) ? void (0) : __assert_fail (
"Exit", "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5438, __extension__
 __PRETTY_FUNCTION__));
assert(isa<BranchInst>(Exit->getTerminator()) &&(static_cast <bool> (isa<BranchInst>(Exit->getTerminator
()) && "Exit block must terminate with unconditional branch"
) ? void (0) : __assert_fail ("isa<BranchInst>(Exit->getTerminator()) && \"Exit block must terminate with unconditional branch\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5440, __extension__
 __PRETTY_FUNCTION__))
       "Exit block must terminate with unconditional branch")(static_cast <bool> (isa<BranchInst>(Exit->getTerminator
()) && "Exit block must terminate with unconditional branch"
) ? void (0) : __assert_fail ("isa<BranchInst>(Exit->getTerminator()) && \"Exit block must terminate with unconditional branch\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5440, __extension__
 __PRETTY_FUNCTION__));
assert(Exit->getSingleSuccessor() == After &&(static_cast <bool> (Exit->getSingleSuccessor() == After
 && "Exit block must jump to after block") ? void (0)
 : __assert_fail ("Exit->getSingleSuccessor() == After && \"Exit block must jump to after block\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5442, __extension__
 __PRETTY_FUNCTION__))
       "Exit block must jump to after block")(static_cast <bool> (Exit->getSingleSuccessor() == After
 && "Exit block must jump to after block") ? void (0)
 : __assert_fail ("Exit->getSingleSuccessor() == After && \"Exit block must jump to after block\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5442, __extension__
 __PRETTY_FUNCTION__));

assert(After)(static_cast <bool> (After) ? void (0) : __assert_fail (
"After", "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5444, __extension__
 __PRETTY_FUNCTION__));
assert(After->getSinglePredecessor() == Exit &&(static_cast <bool> (After->getSinglePredecessor() ==
 Exit && "After block only reachable from exit block"
) ? void (0) : __assert_fail ("After->getSinglePredecessor() == Exit && \"After block only reachable from exit block\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5446, __extension__
 __PRETTY_FUNCTION__))
       "After block only reachable from exit block")(static_cast <bool> (After->getSinglePredecessor() ==
 Exit && "After block only reachable from exit block"
) ? void (0) : __assert_fail ("After->getSinglePredecessor() == Exit && \"After block only reachable from exit block\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5446, __extension__
 __PRETTY_FUNCTION__));
assert(After->empty() || !isa<PHINode>(After->front()))(static_cast <bool> (After->empty() || !isa<PHINode
>(After->front())) ? void (0) : __assert_fail ("After->empty() || !isa<PHINode>(After->front())"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5447, __extension__
 __PRETTY_FUNCTION__));

Instruction *IndVar = getIndVar();
assert(IndVar && "Canonical induction variable not found?")(static_cast <bool> (IndVar && "Canonical induction variable not found?"
) ? void (0) : __assert_fail ("IndVar && \"Canonical induction variable not found?\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5450, __extension__
 __PRETTY_FUNCTION__));
assert(isa<IntegerType>(IndVar->getType()) &&(static_cast <bool> (isa<IntegerType>(IndVar->
getType()) && "Induction variable must be an integer"
) ? void (0) : __assert_fail ("isa<IntegerType>(IndVar->getType()) && \"Induction variable must be an integer\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5452, __extension__
 __PRETTY_FUNCTION__))
       "Induction variable must be an integer")(static_cast <bool> (isa<IntegerType>(IndVar->
getType()) && "Induction variable must be an integer"
) ? void (0) : __assert_fail ("isa<IntegerType>(IndVar->getType()) && \"Induction variable must be an integer\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5452, __extension__
 __PRETTY_FUNCTION__));
assert(cast<PHINode>(IndVar)->getParent() == Header &&(static_cast <bool> (cast<PHINode>(IndVar)->getParent
() == Header && "Induction variable must be a PHI in the loop header"
) ? void (0) : __assert_fail ("cast<PHINode>(IndVar)->getParent() == Header && \"Induction variable must be a PHI in the loop header\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5454, __extension__
 __PRETTY_FUNCTION__))
       "Induction variable must be a PHI in the loop header")(static_cast <bool> (cast<PHINode>(IndVar)->getParent
() == Header && "Induction variable must be a PHI in the loop header"
) ? void (0) : __assert_fail ("cast<PHINode>(IndVar)->getParent() == Header && \"Induction variable must be a PHI in the loop header\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5454, __extension__
 __PRETTY_FUNCTION__));
assert(cast<PHINode>(IndVar)->getIncomingBlock(0) == Preheader)(static_cast <bool> (cast<PHINode>(IndVar)->getIncomingBlock
(0) == Preheader) ? void (0) : __assert_fail ("cast<PHINode>(IndVar)->getIncomingBlock(0) == Preheader"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5455, __extension__
 __PRETTY_FUNCTION__));
assert((static_cast <bool> (cast<ConstantInt>(cast<PHINode
>(IndVar)->getIncomingValue(0))->isZero()) ? void (0
) : __assert_fail ("cast<ConstantInt>(cast<PHINode>(IndVar)->getIncomingValue(0))->isZero()"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5457, __extension__
 __PRETTY_FUNCTION__))
    cast<ConstantInt>(cast<PHINode>(IndVar)->getIncomingValue(0))->isZero())(static_cast <bool> (cast<ConstantInt>(cast<PHINode
>(IndVar)->getIncomingValue(0))->isZero()) ? void (0
) : __assert_fail ("cast<ConstantInt>(cast<PHINode>(IndVar)->getIncomingValue(0))->isZero()"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5457, __extension__
 __PRETTY_FUNCTION__));
assert(cast<PHINode>(IndVar)->getIncomingBlock(1) == Latch)(static_cast <bool> (cast<PHINode>(IndVar)->getIncomingBlock
(1) == Latch) ? void (0) : __assert_fail ("cast<PHINode>(IndVar)->getIncomingBlock(1) == Latch"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5458, __extension__
 __PRETTY_FUNCTION__));

auto *NextIndVar = cast<PHINode>(IndVar)->getIncomingValue(1);
assert(cast<Instruction>(NextIndVar)->getParent() == Latch)(static_cast <bool> (cast<Instruction>(NextIndVar
)->getParent() == Latch) ? void (0) : __assert_fail ("cast<Instruction>(NextIndVar)->getParent() == Latch"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5461, __extension__
 __PRETTY_FUNCTION__));
assert(cast<BinaryOperator>(NextIndVar)->getOpcode() == BinaryOperator::Add)(static_cast <bool> (cast<BinaryOperator>(NextIndVar
)->getOpcode() == BinaryOperator::Add) ? void (0) : __assert_fail
 ("cast<BinaryOperator>(NextIndVar)->getOpcode() == BinaryOperator::Add"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5462, __extension__
 __PRETTY_FUNCTION__));
assert(cast<BinaryOperator>(NextIndVar)->getOperand(0) == IndVar)(static_cast <bool> (cast<BinaryOperator>(NextIndVar
)->getOperand(0) == IndVar) ? void (0) : __assert_fail ("cast<BinaryOperator>(NextIndVar)->getOperand(0) == IndVar"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5463, __extension__
 __PRETTY_FUNCTION__));
assert(cast<ConstantInt>(cast<BinaryOperator>(NextIndVar)->getOperand(1))(static_cast <bool> (cast<ConstantInt>(cast<BinaryOperator
>(NextIndVar)->getOperand(1)) ->isOne()) ? void (0) :
 __assert_fail ("cast<ConstantInt>(cast<BinaryOperator>(NextIndVar)->getOperand(1)) ->isOne()"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5465, __extension__
 __PRETTY_FUNCTION__))
           ->isOne())(static_cast <bool> (cast<ConstantInt>(cast<BinaryOperator
>(NextIndVar)->getOperand(1)) ->isOne()) ? void (0) :
 __assert_fail ("cast<ConstantInt>(cast<BinaryOperator>(NextIndVar)->getOperand(1)) ->isOne()"
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5465, __extension__
 __PRETTY_FUNCTION__));

Value *TripCount = getTripCount();
assert(TripCount && "Loop trip count not found?")(static_cast <bool> (TripCount && "Loop trip count not found?"
) ? void (0) : __assert_fail ("TripCount && \"Loop trip count not found?\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5468, __extension__
 __PRETTY_FUNCTION__));
assert(IndVar->getType() == TripCount->getType() &&(static_cast <bool> (IndVar->getType() == TripCount->
getType() && "Trip count and induction variable must have the same type"
) ? void (0) : __assert_fail ("IndVar->getType() == TripCount->getType() && \"Trip count and induction variable must have the same type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5470, __extension__
 __PRETTY_FUNCTION__))
       "Trip count and induction variable must have the same type")(static_cast <bool> (IndVar->getType() == TripCount->
getType() && "Trip count and induction variable must have the same type"
) ? void (0) : __assert_fail ("IndVar->getType() == TripCount->getType() && \"Trip count and induction variable must have the same type\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5470, __extension__
 __PRETTY_FUNCTION__));

auto *CmpI = cast<CmpInst>(&Cond->front());
assert(CmpI->getPredicate() == CmpInst::ICMP_ULT &&(static_cast <bool> (CmpI->getPredicate() == CmpInst
::ICMP_ULT && "Exit condition must be a signed less-than comparison"
) ? void (0) : __assert_fail ("CmpI->getPredicate() == CmpInst::ICMP_ULT && \"Exit condition must be a signed less-than comparison\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5474, __extension__
 __PRETTY_FUNCTION__))
       "Exit condition must be a signed less-than comparison")(static_cast <bool> (CmpI->getPredicate() == CmpInst
::ICMP_ULT && "Exit condition must be a signed less-than comparison"
) ? void (0) : __assert_fail ("CmpI->getPredicate() == CmpInst::ICMP_ULT && \"Exit condition must be a signed less-than comparison\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5474, __extension__
 __PRETTY_FUNCTION__));
assert(CmpI->getOperand(0) == IndVar &&(static_cast <bool> (CmpI->getOperand(0) == IndVar &&
 "Exit condition must compare the induction variable") ? void
 (0) : __assert_fail ("CmpI->getOperand(0) == IndVar && \"Exit condition must compare the induction variable\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5476, __extension__
 __PRETTY_FUNCTION__))
       "Exit condition must compare the induction variable")(static_cast <bool> (CmpI->getOperand(0) == IndVar &&
 "Exit condition must compare the induction variable") ? void
 (0) : __assert_fail ("CmpI->getOperand(0) == IndVar && \"Exit condition must compare the induction variable\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5476, __extension__
 __PRETTY_FUNCTION__));
assert(CmpI->getOperand(1) == TripCount &&(static_cast <bool> (CmpI->getOperand(1) == TripCount
 && "Exit condition must compare with the trip count"
) ? void (0) : __assert_fail ("CmpI->getOperand(1) == TripCount && \"Exit condition must compare with the trip count\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5478, __extension__
 __PRETTY_FUNCTION__))
       "Exit condition must compare with the trip count")(static_cast <bool> (CmpI->getOperand(1) == TripCount
 && "Exit condition must compare with the trip count"
) ? void (0) : __assert_fail ("CmpI->getOperand(1) == TripCount && \"Exit condition must compare with the trip count\""
, "llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp", 5478, __extension__
 __PRETTY_FUNCTION__));
5479#endif
5480}

5482void CanonicalLoopInfo::invalidate() {
Header = nullptr;
Cond = nullptr;
Latch = nullptr;
Exit = nullptr;
5487}