Cold-Air-Networks/mirror-llvm-project
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 7

Revert "[OpenMP] Restore loop variable values after loop-transformation constructs. (#212853)" (#213127)

Browse source 
Revert "[OpenMP] Restore loop variable values after loop-transformation
constructs. (#212853)"

This reverts commit 2d697f8562.
 Reverting this PR. Bot is failing with:
    Failed Tests (2):
      libomptarget :: spirv64-intel :: offloading/target-tile.c
libomptarget :: x86_64-unknown-linux-gnu :: offloading/target-tile.c
This commit is contained in:
Zahira Ammarguellat 2026-07-30 13:49:00 -07:00 committed by GitHub
parent b05a5d0954
commit f0da3dd928
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
10 changed files with 534 additions and 1221 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

67
clang/include/clang/AST/StmtOpenMP.h
View file

@ -5601,7 +5601,6 @@ class OMPTileDirective final
enum {
PreInitsOffset = 0,
TransformedStmtOffset,
FinalsOffset,
};
explicit OMPTileDirective(SourceLocation StartLoc, SourceLocation EndLoc,
@ -5618,8 +5617,6 @@ class OMPTileDirective final
Data->getChildren()[TransformedStmtOffset] = S;
}
void setFinals(Stmt *Finals) { Data->getChildren()[FinalsOffset] = Finals; }
public:
/// Create a new AST node representation for '#pragma omp tile'.
///
@ -5633,11 +5630,11 @@ public:
/// \param TransformedStmt The loop nest after tiling, or nullptr in
/// dependent contexts.
/// \param PreInits Helper preinits statements for the loop nest.
/// \param Finals Loop variable finalization statements.
static OMPTileDirective *
Create(const ASTContext &C, SourceLocation StartLoc, SourceLocation EndLoc,
ArrayRef<OMPClause *> Clauses, unsigned NumLoops, Stmt *AssociatedStmt,
Stmt *TransformedStmt, Stmt *PreInits, Stmt *Finals);
static OMPTileDirective *Create(const ASTContext &C, SourceLocation StartLoc,
SourceLocation EndLoc,
ArrayRef<OMPClause *> Clauses,
unsigned NumLoops, Stmt *AssociatedStmt,
Stmt *TransformedStmt, Stmt *PreInits);
/// Build an empty '#pragma omp tile' AST node for deserialization.
///
@ -5665,9 +5662,6 @@ public:
/// Return preinits statement.
Stmt *getPreInits() const { return Data->getChildren()[PreInitsOffset]; }
/// Return finals statement (loop variable finalization).
Stmt *getFinals() const { return Data->getChildren()[FinalsOffset]; }
static bool classof(const Stmt *T) {
return T->getStmtClass() == OMPTileDirectiveClass;
}
@ -5683,7 +5677,6 @@ class OMPStripeDirective final
enum {
PreInitsOffset = 0,
TransformedStmtOffset,
FinalsOffset,
};
explicit OMPStripeDirective(SourceLocation StartLoc, SourceLocation EndLoc,
@ -5700,8 +5693,6 @@ class OMPStripeDirective final
Data->getChildren()[TransformedStmtOffset] = S;
}
void setFinals(Stmt *Finals) { Data->getChildren()[FinalsOffset] = Finals; }
public:
/// Create a new AST node representation for '#pragma omp stripe'.
///
@ -5715,11 +5706,10 @@ public:
/// \param TransformedStmt The loop nest after striping, or nullptr in
/// dependent contexts.
/// \param PreInits Helper preinits statements for the loop nest.
/// \param Finals Loop variable finalization statements.
static OMPStripeDirective *
Create(const ASTContext &C, SourceLocation StartLoc, SourceLocation EndLoc,
ArrayRef<OMPClause *> Clauses, unsigned NumLoops, Stmt *AssociatedStmt,
Stmt *TransformedStmt, Stmt *PreInits, Stmt *Finals);
Stmt *TransformedStmt, Stmt *PreInits);
/// Build an empty '#pragma omp stripe' AST node for deserialization.
///
@ -5746,9 +5736,6 @@ public:
/// Return preinits statement.
Stmt *getPreInits() const { return Data->getChildren()[PreInitsOffset]; }
/// Return finals statement (loop variable finalization).
Stmt *getFinals() const { return Data->getChildren()[FinalsOffset]; }
static bool classof(const Stmt *T) {
return T->getStmtClass() == OMPStripeDirectiveClass;
}
@ -5845,7 +5832,6 @@ class OMPReverseDirective final
enum {
PreInitsOffset = 0,
TransformedStmtOffset,
FinalsOffset,
};
explicit OMPReverseDirective(SourceLocation StartLoc, SourceLocation EndLoc,
@ -5862,8 +5848,6 @@ class OMPReverseDirective final
Data->getChildren()[TransformedStmtOffset] = S;
}
void setFinals(Stmt *Finals) { Data->getChildren()[FinalsOffset] = Finals; }
public:
/// Create a new AST node representation for '#pragma omp reverse'.
///
@ -5875,11 +5859,11 @@ public:
/// \param TransformedStmt The loop nest after tiling, or nullptr in
/// dependent contexts.
/// \param PreInits Helper preinits statements for the loop nest.
/// \param Finals Loop variable finalization statements.
static OMPReverseDirective *
Create(const ASTContext &C, SourceLocation StartLoc, SourceLocation EndLoc,
Stmt *AssociatedStmt, unsigned NumLoops, Stmt *TransformedStmt,
Stmt *PreInits, Stmt *Finals);
static OMPReverseDirective *Create(const ASTContext &C,
SourceLocation StartLoc,
SourceLocation EndLoc,
Stmt *AssociatedStmt, unsigned NumLoops,
Stmt *TransformedStmt, Stmt *PreInits);
/// Build an empty '#pragma omp reverse' AST node for deserialization.
///
@ -5897,9 +5881,6 @@ public:
/// Return preinits statement.
Stmt *getPreInits() const { return Data->getChildren()[PreInitsOffset]; }
/// Return finals statement.
Stmt *getFinals() const { return Data->getChildren()[FinalsOffset]; }
static bool classof(const Stmt *T) {
return T->getStmtClass() == OMPReverseDirectiveClass;
}
@ -5922,7 +5903,6 @@ class OMPInterchangeDirective final
enum {
PreInitsOffset = 0,
TransformedStmtOffset,
FinalsOffset,
};
explicit OMPInterchangeDirective(SourceLocation StartLoc,
@ -5939,8 +5919,6 @@ class OMPInterchangeDirective final
Data->getChildren()[TransformedStmtOffset] = S;
}
void setFinals(Stmt *Finals) { Data->getChildren()[FinalsOffset] = Finals; }
public:
/// Create a new AST node representation for '#pragma omp interchange'.
///
@ -5954,11 +5932,10 @@ public:
/// \param TransformedStmt The loop nest after tiling, or nullptr in
/// dependent contexts.
/// \param PreInits Helper preinits statements for the loop nest.
/// \param Finals Loop variable finalization statements.
static OMPInterchangeDirective *
Create(const ASTContext &C, SourceLocation StartLoc, SourceLocation EndLoc,
ArrayRef<OMPClause *> Clauses, unsigned NumLoops, Stmt *AssociatedStmt,
Stmt *TransformedStmt, Stmt *PreInits, Stmt *Finals);
Stmt *TransformedStmt, Stmt *PreInits);
/// Build an empty '#pragma omp interchange' AST node for deserialization.
///
@ -5977,9 +5954,6 @@ public:
/// Return preinits statement.
Stmt *getPreInits() const { return Data->getChildren()[PreInitsOffset]; }
/// Return finals statement.
Stmt *getFinals() const { return Data->getChildren()[FinalsOffset]; }
static bool classof(const Stmt *T) {
return T->getStmtClass() == OMPInterchangeDirectiveClass;
}
@ -6035,7 +6009,6 @@ class OMPFuseDirective final
enum {
PreInitsOffset = 0,
TransformedStmtOffset,
FinalsOffset,
};
explicit OMPFuseDirective(SourceLocation StartLoc, SourceLocation EndLoc)
@ -6050,8 +6023,6 @@ class OMPFuseDirective final
Data->getChildren()[TransformedStmtOffset] = S;
}
void setFinals(Stmt *Finals) { Data->getChildren()[FinalsOffset] = Finals; }
public:
/// Create a new AST node representation for #pragma omp fuse'
///
@ -6067,13 +6038,10 @@ public:
/// \param TransformedStmt The loop nest after fusion, or nullptr in
/// dependent
/// \param PreInits Helper preinits statements for the loop nest
/// \param Finals Loop variable finalization statements
static OMPFuseDirective *Create(const ASTContext &C, SourceLocation StartLoc,
SourceLocation EndLoc,
ArrayRef<OMPClause *> Clauses,
unsigned NumGeneratedTopLevelLoops,
Stmt *AssociatedStmt, Stmt *TransformedStmt,
Stmt *PreInits, Stmt *Finals);
static OMPFuseDirective *
Create(const ASTContext &C, SourceLocation StartLoc, SourceLocation EndLoc,
ArrayRef<OMPClause *> Clauses, unsigned NumGeneratedTopLevelLoops,
Stmt *AssociatedStmt, Stmt *TransformedStmt, Stmt *PreInits);
/// Build an empty '#pragma omp fuse' AST node for deserialization
///
@ -6092,9 +6060,6 @@ public:
/// Return preinits statement.
Stmt *getPreInits() const { return Data->getChildren()[PreInitsOffset]; }
/// Return finals statement.
Stmt *getFinals() const { return Data->getChildren()[FinalsOffset]; }
static bool classof(const Stmt *T) {
return T->getStmtClass() == OMPFuseDirectiveClass;
}

48
clang/lib/AST/StmtOpenMP.cpp
View file

@ -452,12 +452,12 @@ OMPTileDirective *
OMPTileDirective::Create(const ASTContext &C, SourceLocation StartLoc,
SourceLocation EndLoc, ArrayRef<OMPClause *> Clauses,
unsigned NumLoops, Stmt *AssociatedStmt,
Stmt *TransformedStmt, Stmt *PreInits, Stmt *Finals) {
Stmt *TransformedStmt, Stmt *PreInits) {
OMPTileDirective *Dir = createDirective<OMPTileDirective>(
C, Clauses, AssociatedStmt, FinalsOffset + 1, StartLoc, EndLoc, NumLoops);
C, Clauses, AssociatedStmt, TransformedStmtOffset + 1, StartLoc, EndLoc,
NumLoops);
Dir->setTransformedStmt(TransformedStmt);
Dir->setPreInits(PreInits);
Dir->setFinals(Finals);
return Dir;
}
@ -465,19 +465,20 @@ OMPTileDirective *OMPTileDirective::CreateEmpty(const ASTContext &C,
unsigned NumClauses,
unsigned NumLoops) {
return createEmptyDirective<OMPTileDirective>(
C, NumClauses, /*HasAssociatedStmt=*/true, FinalsOffset + 1,
C, NumClauses, /*HasAssociatedStmt=*/true, TransformedStmtOffset + 1,
SourceLocation(), SourceLocation(), NumLoops);
}
OMPStripeDirective *OMPStripeDirective::Create(
const ASTContext &C, SourceLocation StartLoc, SourceLocation EndLoc,
ArrayRef<OMPClause *> Clauses, unsigned NumLoops, Stmt *AssociatedStmt,
Stmt *TransformedStmt, Stmt *PreInits, Stmt *Finals) {
OMPStripeDirective *
OMPStripeDirective::Create(const ASTContext &C, SourceLocation StartLoc,
SourceLocation EndLoc, ArrayRef<OMPClause *> Clauses,
unsigned NumLoops, Stmt *AssociatedStmt,
Stmt *TransformedStmt, Stmt *PreInits) {
OMPStripeDirective *Dir = createDirective<OMPStripeDirective>(
C, Clauses, AssociatedStmt, FinalsOffset + 1, StartLoc, EndLoc, NumLoops);
C, Clauses, AssociatedStmt, TransformedStmtOffset + 1, StartLoc, EndLoc,
NumLoops);
Dir->setTransformedStmt(TransformedStmt);
Dir->setPreInits(PreInits);
Dir->setFinals(Finals);
return Dir;
}
@ -485,7 +486,7 @@ OMPStripeDirective *OMPStripeDirective::CreateEmpty(const ASTContext &C,
unsigned NumClauses,
unsigned NumLoops) {
return createEmptyDirective<OMPStripeDirective>(
C, NumClauses, /*HasAssociatedStmt=*/true, FinalsOffset + 1,
C, NumClauses, /*HasAssociatedStmt=*/true, TransformedStmtOffset + 1,
SourceLocation(), SourceLocation(), NumLoops);
}
@ -515,31 +516,31 @@ OMPReverseDirective *
OMPReverseDirective::Create(const ASTContext &C, SourceLocation StartLoc,
SourceLocation EndLoc, Stmt *AssociatedStmt,
unsigned NumLoops, Stmt *TransformedStmt,
Stmt *PreInits, Stmt *Finals) {
Stmt *PreInits) {
OMPReverseDirective *Dir = createDirective<OMPReverseDirective>(
C, {}, AssociatedStmt, FinalsOffset + 1, StartLoc, EndLoc, NumLoops);
C, {}, AssociatedStmt, TransformedStmtOffset + 1, StartLoc, EndLoc,
NumLoops);
Dir->setTransformedStmt(TransformedStmt);
Dir->setPreInits(PreInits);
Dir->setFinals(Finals);
return Dir;
}
OMPReverseDirective *OMPReverseDirective::CreateEmpty(const ASTContext &C,
unsigned NumLoops) {
return createEmptyDirective<OMPReverseDirective>(
C, /*NumClauses=*/0, /*HasAssociatedStmt=*/true, FinalsOffset + 1,
SourceLocation(), SourceLocation(), NumLoops);
C, /*NumClauses=*/0, /*HasAssociatedStmt=*/true,
TransformedStmtOffset + 1, SourceLocation(), SourceLocation(), NumLoops);
}
OMPInterchangeDirective *OMPInterchangeDirective::Create(
const ASTContext &C, SourceLocation StartLoc, SourceLocation EndLoc,
ArrayRef<OMPClause *> Clauses, unsigned NumLoops, Stmt *AssociatedStmt,
Stmt *TransformedStmt, Stmt *PreInits, Stmt *Finals) {
Stmt *TransformedStmt, Stmt *PreInits) {
OMPInterchangeDirective *Dir = createDirective<OMPInterchangeDirective>(
C, Clauses, AssociatedStmt, FinalsOffset + 1, StartLoc, EndLoc, NumLoops);
C, Clauses, AssociatedStmt, TransformedStmtOffset + 1, StartLoc, EndLoc,
NumLoops);
Dir->setTransformedStmt(TransformedStmt);
Dir->setPreInits(PreInits);
Dir->setFinals(Finals);
return Dir;
}
@ -547,7 +548,7 @@ OMPInterchangeDirective *
OMPInterchangeDirective::CreateEmpty(const ASTContext &C, unsigned NumClauses,
unsigned NumLoops) {
return createEmptyDirective<OMPInterchangeDirective>(
C, NumClauses, /*HasAssociatedStmt=*/true, FinalsOffset + 1,
C, NumClauses, /*HasAssociatedStmt=*/true, TransformedStmtOffset + 1,
SourceLocation(), SourceLocation(), NumLoops);
}
@ -575,13 +576,12 @@ OMPSplitDirective *OMPSplitDirective::CreateEmpty(const ASTContext &C,
OMPFuseDirective *OMPFuseDirective::Create(
const ASTContext &C, SourceLocation StartLoc, SourceLocation EndLoc,
ArrayRef<OMPClause *> Clauses, unsigned NumGeneratedTopLevelLoops,
Stmt *AssociatedStmt, Stmt *TransformedStmt, Stmt *PreInits, Stmt *Finals) {
Stmt *AssociatedStmt, Stmt *TransformedStmt, Stmt *PreInits) {
OMPFuseDirective *Dir = createDirective<OMPFuseDirective>(
C, Clauses, AssociatedStmt, FinalsOffset + 1, StartLoc, EndLoc);
C, Clauses, AssociatedStmt, TransformedStmtOffset + 1, StartLoc, EndLoc);
Dir->setTransformedStmt(TransformedStmt);
Dir->setPreInits(PreInits);
Dir->setFinals(Finals);
Dir->setNumGeneratedTopLevelLoops(NumGeneratedTopLevelLoops);
return Dir;
}
@ -589,7 +589,7 @@ OMPFuseDirective *OMPFuseDirective::Create(
OMPFuseDirective *OMPFuseDirective::CreateEmpty(const ASTContext &C,
unsigned NumClauses) {
OMPFuseDirective *Dir = createEmptyDirective<OMPFuseDirective>(
C, NumClauses, /*HasAssociatedStmt=*/true, FinalsOffset + 1,
C, NumClauses, /*HasAssociatedStmt=*/true, TransformedStmtOffset + 1,
SourceLocation(), SourceLocation());
return Dir;
}

25
clang/lib/CodeGen/CGStmtOpenMP.cpp
View file

@ -3221,33 +3221,18 @@ void CodeGenFunction::EmitOMPTileDirective(const OMPTileDirective &S) {
// Emit the de-sugared statement.
OMPTransformDirectiveScopeRAII TileScope(*this, &S);
EmitStmt(S.getTransformedStmt());
// Emit loop variable finalization as required by OpenMP 6.0 spec to restore
// original loop variable values after the loop-transformation construct.
if (auto *Finals = S.getFinals())
EmitStmt(Finals);
}
void CodeGenFunction::EmitOMPStripeDirective(const OMPStripeDirective &S) {
// Emit the de-sugared statement.
OMPTransformDirectiveScopeRAII StripeScope(*this, &S);
EmitStmt(S.getTransformedStmt());
// Emit loop variable finalization as required by OpenMP 6.0 spec to restore
// original loop variable values after the loop-transformation construct.
if (auto *Finals = S.getFinals())
EmitStmt(Finals);
}
void CodeGenFunction::EmitOMPReverseDirective(const OMPReverseDirective &S) {
// Emit the de-sugared statement.
OMPTransformDirectiveScopeRAII ReverseScope(*this, &S);
EmitStmt(S.getTransformedStmt());
// Emit loop variable finalization as required by OpenMP 6.0 spec to restore
// original loop variable values after the loop-transformation construct.
if (auto *Finals = S.getFinals())
EmitStmt(Finals);
}
void CodeGenFunction::EmitOMPSplitDirective(const OMPSplitDirective &S) {
@ -3261,22 +3246,12 @@ void CodeGenFunction::EmitOMPInterchangeDirective(
// Emit the de-sugared statement.
OMPTransformDirectiveScopeRAII InterchangeScope(*this, &S);
EmitStmt(S.getTransformedStmt());
// Emit loop variable finalization as required by OpenMP 6.0 spec to restore
// original loop variable values after the loop-transformation construct.
if (auto *Finals = S.getFinals())
EmitStmt(Finals);
}
void CodeGenFunction::EmitOMPFuseDirective(const OMPFuseDirective &S) {
// Emit the de-sugared statement
OMPTransformDirectiveScopeRAII FuseScope(*this, &S);
EmitStmt(S.getTransformedStmt());
// Emit loop variable finalization as required by OpenMP 6.0 spec to restore
// original loop variable values after the loop-transformation construct.
if (auto *Finals = S.getFinals())
EmitStmt(Finals);
}
void CodeGenFunction::EmitOMPUnrollDirective(const OMPUnrollDirective &S) {

111
clang/lib/Sema/SemaOpenMP.cpp
View file

@ -15033,48 +15033,6 @@ static Expr *makeFloorIVRef(Sema &SemaRef, ArrayRef<VarDecl *> FloorIndVars,
OrigCntVar->getExprLoc());
}
/// Build loop variable finalization statement from HelperExprs.Finals.
/// Returns a CompoundStmt containing all finalization statements, or nullptr
/// if there are no finalization statements.
/// Note: Loops with non-arithmetic loop variables (e.g., iterators) are skipped
/// because finalization only applies to integer/floating-point counters.
static Stmt *
buildLoopFinalization(ASTContext &Context,
ArrayRef<OMPLoopBasedDirective::HelperExprs> LoopHelpers,
ArrayRef<Stmt *> LoopStmts) {
assert(LoopHelpers.size() == LoopStmts.size() &&
"LoopHelpers and LoopStmts must have the same size");
SmallVector<Stmt *, 8> FinalizationStmts;
for (const auto &[Helper, LoopStmt] : llvm::zip(LoopHelpers, LoopStmts)) {
// Skip finalization for range-based for loops (CXXForRangeStmt) since
// their loop variables are iterators, not integer counters.
if (isa<CXXForRangeStmt>(LoopStmt))
continue;
// For ForStmt, check if the loop variable is arithmetic.
if (auto *For = dyn_cast<ForStmt>(LoopStmt)) {
if (auto *InitDeclStmt = dyn_cast_or_null<DeclStmt>(For->getInit())) {
if (InitDeclStmt->isSingleDecl()) {
if (auto *InitVar =
dyn_cast<VarDecl>(InitDeclStmt->getSingleDecl())) {
// Skip finalization if loop variable is not arithmetic.
if (!InitVar->getType()->isArithmeticType())
continue;
}
}
}
}
for (auto *Final : Helper.Finals)
if (Final)
FinalizationStmts.push_back(Final);
}
if (FinalizationStmts.empty())
return nullptr;
return CompoundStmt::Create(Context, FinalizationStmts, FPOptionsOverride(),
SourceLocation(), SourceLocation());
}
StmtResult SemaOpenMP::ActOnOpenMPTileDirective(ArrayRef<OMPClause *> Clauses,
Stmt *AStmt,
SourceLocation StartLoc,
@ -15104,7 +15062,7 @@ StmtResult SemaOpenMP::ActOnOpenMPTileDirective(ArrayRef<OMPClause *> Clauses,
// Delay tiling to when template is completely instantiated.
if (SemaRef.CurContext->isDependentContext())
return OMPTileDirective::Create(Context, StartLoc, EndLoc, Clauses,
NumLoops, AStmt, nullptr, nullptr, nullptr);
NumLoops, AStmt, nullptr, nullptr);
assert(LoopHelpers.size() == NumLoops &&
"Expecting loop iteration space dimensionality to match number of "
@ -15345,10 +15303,9 @@ StmtResult SemaOpenMP::ActOnOpenMPTileDirective(ArrayRef<OMPClause *> Clauses,
LoopHelper.Init->getBeginLoc(), LoopHelper.Inc->getEndLoc());
}
return OMPTileDirective::Create(
Context, StartLoc, EndLoc, Clauses, NumLoops, AStmt, Inner,
buildPreInits(Context, PreInits),
buildLoopFinalization(Context, LoopHelpers, LoopStmts));
return OMPTileDirective::Create(Context, StartLoc, EndLoc, Clauses, NumLoops,
AStmt, Inner,
buildPreInits(Context, PreInits));
}
StmtResult SemaOpenMP::ActOnOpenMPStripeDirective(ArrayRef<OMPClause *> Clauses,
@ -15382,8 +15339,7 @@ StmtResult SemaOpenMP::ActOnOpenMPStripeDirective(ArrayRef<OMPClause *> Clauses,
// Delay striping to when template is completely instantiated.
if (SemaRef.CurContext->isDependentContext())
return OMPStripeDirective::Create(Context, StartLoc, EndLoc, Clauses,
NumLoops, AStmt, nullptr, nullptr,
nullptr);
NumLoops, AStmt, nullptr, nullptr);
assert(LoopHelpers.size() == NumLoops &&
"Expecting loop iteration space dimensionality to match number of "
@ -15606,10 +15562,9 @@ StmtResult SemaOpenMP::ActOnOpenMPStripeDirective(ArrayRef<OMPClause *> Clauses,
LoopHelper.Init->getBeginLoc(), LoopHelper.Inc->getEndLoc());
}
return OMPStripeDirective::Create(
Context, StartLoc, EndLoc, Clauses, NumLoops, AStmt, Inner,
buildPreInits(Context, PreInits),
buildLoopFinalization(Context, LoopHelpers, LoopStmts));
return OMPStripeDirective::Create(Context, StartLoc, EndLoc, Clauses,
NumLoops, AStmt, Inner,
buildPreInits(Context, PreInits));
}
StmtResult SemaOpenMP::ActOnOpenMPUnrollDirective(ArrayRef<OMPClause *> Clauses,
@ -15916,7 +15871,7 @@ StmtResult SemaOpenMP::ActOnOpenMPReverseDirective(Stmt *AStmt,
// instantiated.
if (SemaRef.CurContext->isDependentContext())
return OMPReverseDirective::Create(Context, StartLoc, EndLoc, AStmt,
NumLoops, nullptr, nullptr, nullptr);
NumLoops, nullptr, nullptr);
assert(LoopHelpers.size() == NumLoops &&
"Expecting a single-dimensional loop iteration space");
@ -16075,11 +16030,9 @@ StmtResult SemaOpenMP::ActOnOpenMPReverseDirective(Stmt *AStmt,
ForStmt(Context, Init.get(), Cond.get(), nullptr, Incr.get(),
ReversedBody, LoopHelper.Init->getBeginLoc(),
LoopHelper.Init->getBeginLoc(), LoopHelper.Inc->getEndLoc());
return OMPReverseDirective::Create(
Context, StartLoc, EndLoc, AStmt, NumLoops, ReversedFor,
buildPreInits(Context, PreInits),
buildLoopFinalization(Context, LoopHelpers, {LoopStmt}));
return OMPReverseDirective::Create(Context, StartLoc, EndLoc, AStmt, NumLoops,
ReversedFor,
buildPreInits(Context, PreInits));
}
/// Build the AST for \#pragma omp split counts(c1, c2, ...).
@ -16338,8 +16291,7 @@ StmtResult SemaOpenMP::ActOnOpenMPInterchangeDirective(
// Delay interchange to when template is completely instantiated.
if (CurContext->isDependentContext())
return OMPInterchangeDirective::Create(Context, StartLoc, EndLoc, Clauses,
NumLoops, AStmt, nullptr, nullptr,
nullptr);
NumLoops, AStmt, nullptr, nullptr);
// An invalid expression in the permutation clause is set to nullptr in
// ActOnOpenMPPermutationClause.
@ -16392,8 +16344,7 @@ StmtResult SemaOpenMP::ActOnOpenMPInterchangeDirective(
return Idx == Arg;
}))
return OMPInterchangeDirective::Create(Context, StartLoc, EndLoc, Clauses,
NumLoops, AStmt, AStmt, nullptr,
nullptr);
NumLoops, AStmt, AStmt, nullptr);
// Find the affected loops.
SmallVector<Stmt *> LoopStmts(NumLoops, nullptr);
@ -16503,10 +16454,9 @@ StmtResult SemaOpenMP::ActOnOpenMPInterchangeDirective(
SourceHelper.Inc->getEndLoc());
}
return OMPInterchangeDirective::Create(
Context, StartLoc, EndLoc, Clauses, NumLoops, AStmt, Inner,
buildPreInits(Context, PreInits),
buildLoopFinalization(Context, LoopHelpers, LoopStmts));
return OMPInterchangeDirective::Create(Context, StartLoc, EndLoc, Clauses,
NumLoops, AStmt, Inner,
buildPreInits(Context, PreInits));
}
StmtResult SemaOpenMP::ActOnOpenMPFuseDirective(ArrayRef<OMPClause *> Clauses,
@ -16529,8 +16479,7 @@ StmtResult SemaOpenMP::ActOnOpenMPFuseDirective(ArrayRef<OMPClause *> Clauses,
// because a dependent context could prevent determining its true value
if (CurrContext->isDependentContext())
return OMPFuseDirective::Create(Context, StartLoc, EndLoc, Clauses,
/* NumLoops */ 1, AStmt, nullptr, nullptr,
nullptr);
/* NumLoops */ 1, AStmt, nullptr, nullptr);
// Validate that the potential loop sequence is transformable for fusion
// Also collect the HelperExprs, Loop Stmts, Inits, and Number of loops
@ -16621,6 +16570,7 @@ StmtResult SemaOpenMP::ActOnOpenMPFuseDirective(ArrayRef<OMPClause *> Clauses,
SmallVector<VarDecl *, 4> LBVarDecls;
SmallVector<VarDecl *, 4> STVarDecls;
SmallVector<VarDecl *, 4> NIVarDecls;
SmallVector<VarDecl *, 4> UBVarDecls;
SmallVector<VarDecl *, 4> IVVarDecls;
// Helper lambda to create variables for bounds, strides, and other
@ -16703,6 +16653,8 @@ StmtResult SemaOpenMP::ActOnOpenMPFuseDirective(ArrayRef<OMPClause *> Clauses,
SeqAnalysis.Loops[I].TheForStmt,
SeqAnalysis.Loops[I].OriginalInits, PreInits);
}
auto [UBVD, UBDStmt] =
CreateHelperVarAndStmt(SeqAnalysis.Loops[I].HelperExprs.UB, "ub", J);
auto [LBVD, LBDStmt] =
CreateHelperVarAndStmt(SeqAnalysis.Loops[I].HelperExprs.LB, "lb", J);
auto [STVD, STDStmt] =
@ -16715,6 +16667,7 @@ StmtResult SemaOpenMP::ActOnOpenMPFuseDirective(ArrayRef<OMPClause *> Clauses,
assert(LBVD && STVD && NIVD && IVVD &&
"OpenMP Fuse Helper variables creation failed");
UBVarDecls.push_back(UBVD);
LBVarDecls.push_back(LBVD);
STVarDecls.push_back(STVD);
NIVarDecls.push_back(NIVD);
@ -16932,18 +16885,6 @@ StmtResult SemaOpenMP::ActOnOpenMPFuseDirective(ArrayRef<OMPClause *> Clauses,
FusedBody, InitStmt.get()->getBeginLoc(), SourceLocation(),
IncrExpr.get()->getEndLoc());
// 8. Build finalization statements for fused loops.
// Collect HelperExprs and loop statements from the fused loops.
SmallVector<OMPLoopBasedDirective::HelperExprs, 4> FusedLoopHelpers;
SmallVector<Stmt *, 4> FusedLoopStmts;
for (unsigned I = FirstVal - 1; I < LastVal; ++I) {
if (SeqAnalysis.Loops[I].isRegularLoop() &&
isa<ForStmt>(SeqAnalysis.Loops[I].TheForStmt)) {
FusedLoopHelpers.push_back(SeqAnalysis.Loops[I].HelperExprs);
FusedLoopStmts.push_back(SeqAnalysis.Loops[I].TheForStmt);
}
}
// In the case of looprange, the result of fuse won't simply
// be a single loop (ForStmt), but rather a loop sequence
// (CompoundStmt) of 3 parts: the pre-fusion loops, the fused loop
@ -16991,11 +16932,9 @@ StmtResult SemaOpenMP::ActOnOpenMPFuseDirective(ArrayRef<OMPClause *> Clauses,
FusionStmt = CompoundStmt::Create(Context, FinalLoops, FPOptionsOverride(),
SourceLocation(), SourceLocation());
}
return OMPFuseDirective::Create(
Context, StartLoc, EndLoc, Clauses, NumGeneratedTopLevelLoops, AStmt,
FusionStmt, buildPreInits(Context, PreInits),
buildLoopFinalization(Context, FusedLoopHelpers, FusedLoopStmts));
return OMPFuseDirective::Create(Context, StartLoc, EndLoc, Clauses,
NumGeneratedTopLevelLoops, AStmt, FusionStmt,
buildPreInits(Context, PreInits));
}
OMPClause *SemaOpenMP::ActOnOpenMPSingleExprClause(OpenMPClauseKind Kind,

282
clang/test/OpenMP/fuse_codegen.cpp
View file

@ -242,34 +242,8 @@ extern "C" void foo5() {
// CHECK1-NEXT: [[TMP43:%.*]] = load i32, ptr [[DOTOMP_FUSE_INDEX]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add i32 [[TMP43]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTOMP_FUSE_INDEX]], align 4
// CHECK1-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP2:![0-9]+]]
// CHECK1-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK1: [[FOR_END]]:
// CHECK1-NEXT: [[TMP44:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[TMP45:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK1-NEXT: [[TMP46:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[SUB28:%.*]] = sub i32 [[TMP45]], [[TMP46]]
// CHECK1-NEXT: [[SUB29:%.*]] = sub i32 [[SUB28]], 1
// CHECK1-NEXT: [[TMP47:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK1-NEXT: [[ADD30:%.*]] = add i32 [[SUB29]], [[TMP47]]
// CHECK1-NEXT: [[TMP48:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK1-NEXT: [[DIV31:%.*]] = udiv i32 [[ADD30]], [[TMP48]]
// CHECK1-NEXT: [[TMP49:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK1-NEXT: [[MUL32:%.*]] = mul i32 [[DIV31]], [[TMP49]]
// CHECK1-NEXT: [[ADD33:%.*]] = add i32 [[TMP44]], [[MUL32]]
// CHECK1-NEXT: store i32 [[ADD33]], ptr [[I]], align 4
// CHECK1-NEXT: [[TMP50:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_6]], align 4
// CHECK1-NEXT: [[TMP51:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_7]], align 4
// CHECK1-NEXT: [[TMP52:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_6]], align 4
// CHECK1-NEXT: [[SUB34:%.*]] = sub i32 [[TMP51]], [[TMP52]]
// CHECK1-NEXT: [[SUB35:%.*]] = sub i32 [[SUB34]], 1
// CHECK1-NEXT: [[TMP53:%.*]] = load i32, ptr [[DOTNEW_STEP8]], align 4
// CHECK1-NEXT: [[ADD36:%.*]] = add i32 [[SUB35]], [[TMP53]]
// CHECK1-NEXT: [[TMP54:%.*]] = load i32, ptr [[DOTNEW_STEP8]], align 4
// CHECK1-NEXT: [[DIV37:%.*]] = udiv i32 [[ADD36]], [[TMP54]]
// CHECK1-NEXT: [[TMP55:%.*]] = load i32, ptr [[DOTNEW_STEP8]], align 4
// CHECK1-NEXT: [[MUL38:%.*]] = mul i32 [[DIV37]], [[TMP55]]
// CHECK1-NEXT: [[ADD39:%.*]] = add i32 [[TMP50]], [[MUL38]]
// CHECK1-NEXT: store i32 [[ADD39]], ptr [[J]], align 4
// CHECK1-NEXT: ret void
//
//
@ -499,47 +473,8 @@ extern "C" void foo5() {
// CHECK1-NEXT: [[TMP68:%.*]] = load i32, ptr [[DOTOMP_FUSE_INDEX]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add i32 [[TMP68]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTOMP_FUSE_INDEX]], align 4
// CHECK1-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]]
// CHECK1-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]]
// CHECK1: [[FOR_END]]:
// CHECK1-NEXT: [[TMP69:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[TMP70:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK1-NEXT: [[TMP71:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[SUB53:%.*]] = sub i32 [[TMP70]], [[TMP71]]
// CHECK1-NEXT: [[SUB54:%.*]] = sub i32 [[SUB53]], 1
// CHECK1-NEXT: [[TMP72:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK1-NEXT: [[ADD55:%.*]] = add i32 [[SUB54]], [[TMP72]]
// CHECK1-NEXT: [[TMP73:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK1-NEXT: [[DIV56:%.*]] = udiv i32 [[ADD55]], [[TMP73]]
// CHECK1-NEXT: [[TMP74:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK1-NEXT: [[MUL57:%.*]] = mul i32 [[DIV56]], [[TMP74]]
// CHECK1-NEXT: [[ADD58:%.*]] = add i32 [[TMP69]], [[MUL57]]
// CHECK1-NEXT: store i32 [[ADD58]], ptr [[I]], align 4
// CHECK1-NEXT: [[TMP75:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_6]], align 4
// CHECK1-NEXT: [[TMP76:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_6]], align 4
// CHECK1-NEXT: [[TMP77:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_7]], align 4
// CHECK1-NEXT: [[SUB59:%.*]] = sub i32 [[TMP76]], [[TMP77]]
// CHECK1-NEXT: [[SUB60:%.*]] = sub i32 [[SUB59]], 1
// CHECK1-NEXT: [[TMP78:%.*]] = load i32, ptr [[DOTNEW_STEP8]], align 4
// CHECK1-NEXT: [[ADD61:%.*]] = add i32 [[SUB60]], [[TMP78]]
// CHECK1-NEXT: [[TMP79:%.*]] = load i32, ptr [[DOTNEW_STEP8]], align 4
// CHECK1-NEXT: [[DIV62:%.*]] = udiv i32 [[ADD61]], [[TMP79]]
// CHECK1-NEXT: [[TMP80:%.*]] = load i32, ptr [[DOTNEW_STEP8]], align 4
// CHECK1-NEXT: [[MUL63:%.*]] = mul i32 [[DIV62]], [[TMP80]]
// CHECK1-NEXT: [[SUB64:%.*]] = sub i32 [[TMP75]], [[MUL63]]
// CHECK1-NEXT: store i32 [[SUB64]], ptr [[J]], align 4
// CHECK1-NEXT: [[TMP81:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_17]], align 4
// CHECK1-NEXT: [[TMP82:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_19]], align 4
// CHECK1-NEXT: [[TMP83:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_17]], align 4
// CHECK1-NEXT: [[SUB65:%.*]] = sub i32 [[TMP82]], [[TMP83]]
// CHECK1-NEXT: [[SUB66:%.*]] = sub i32 [[SUB65]], 1
// CHECK1-NEXT: [[TMP84:%.*]] = load i32, ptr [[DOTNEW_STEP21]], align 4
// CHECK1-NEXT: [[ADD67:%.*]] = add i32 [[SUB66]], [[TMP84]]
// CHECK1-NEXT: [[TMP85:%.*]] = load i32, ptr [[DOTNEW_STEP21]], align 4
// CHECK1-NEXT: [[DIV68:%.*]] = udiv i32 [[ADD67]], [[TMP85]]
// CHECK1-NEXT: [[TMP86:%.*]] = load i32, ptr [[DOTNEW_STEP21]], align 4
// CHECK1-NEXT: [[MUL69:%.*]] = mul i32 [[DIV68]], [[TMP86]]
// CHECK1-NEXT: [[ADD70:%.*]] = add i32 [[TMP81]], [[MUL69]]
// CHECK1-NEXT: store i32 [[ADD70]], ptr [[K]], align 4
// CHECK1-NEXT: ret void
//
//
@ -633,14 +568,14 @@ extern "C" void foo5() {
// CHECK1-NEXT: store i64 [[CONV]], ptr [[DOTOMP_NI05]], align 8
// CHECK1-NEXT: store i32 42, ptr [[C]], align 4
// CHECK1-NEXT: store ptr [[ARR]], ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META5:![0-9]+]], !align [[META6:![0-9]+]]
// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP8]], i64 0, i64 0
// CHECK1-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 256
// CHECK1-NEXT: store ptr [[ADD_PTR]], ptr [[__END2]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META5]], !align [[META6]]
// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY7:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP9]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY7]], ptr [[__BEGIN2]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META5]], !align [[META6]]
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY9:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP10]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY9]], ptr [[DOTCAPTURE_EXPR_8]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[__END2]], align 8
@ -663,14 +598,14 @@ extern "C" void foo5() {
// CHECK1-NEXT: store i64 [[ADD19]], ptr [[DOTOMP_NI118]], align 8
// CHECK1-NEXT: store i32 37, ptr [[CC]], align 4
// CHECK1-NEXT: store ptr [[ARR]], ptr [[__RANGE221]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = load ptr, ptr [[__RANGE221]], align 8, !nonnull [[META5]], !align [[META6]]
// CHECK1-NEXT: [[TMP15:%.*]] = load ptr, ptr [[__RANGE221]], align 8
// CHECK1-NEXT: [[ARRAYDECAY23:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP15]], i64 0, i64 0
// CHECK1-NEXT: [[ADD_PTR24:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY23]], i64 256
// CHECK1-NEXT: store ptr [[ADD_PTR24]], ptr [[__END222]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = load ptr, ptr [[__RANGE221]], align 8, !nonnull [[META5]], !align [[META6]]
// CHECK1-NEXT: [[TMP16:%.*]] = load ptr, ptr [[__RANGE221]], align 8
// CHECK1-NEXT: [[ARRAYDECAY26:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP16]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY26]], ptr [[__BEGIN225]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = load ptr, ptr [[__RANGE221]], align 8, !nonnull [[META5]], !align [[META6]]
// CHECK1-NEXT: [[TMP17:%.*]] = load ptr, ptr [[__RANGE221]], align 8
// CHECK1-NEXT: [[ARRAYDECAY28:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP17]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY28]], ptr [[DOTCAPTURE_EXPR_27]], align 8
// CHECK1-NEXT: [[TMP18:%.*]] = load ptr, ptr [[__END222]], align 8
@ -804,7 +739,7 @@ extern "C" void foo5() {
// CHECK1-NEXT: [[TMP60:%.*]] = load ptr, ptr [[__BEGIN2]], align 8
// CHECK1-NEXT: store ptr [[TMP60]], ptr [[V]], align 8
// CHECK1-NEXT: [[TMP61:%.*]] = load i32, ptr [[C]], align 4
// CHECK1-NEXT: [[TMP62:%.*]] = load ptr, ptr [[V]], align 8, !nonnull [[META5]], !align [[META6]]
// CHECK1-NEXT: [[TMP62:%.*]] = load ptr, ptr [[V]], align 8
// CHECK1-NEXT: [[TMP63:%.*]] = load double, ptr [[TMP62]], align 8
// CHECK1-NEXT: call void (...) @body(i32 noundef [[TMP61]], double noundef [[TMP63]])
// CHECK1-NEXT: br label %[[IF_END81]]
@ -828,7 +763,7 @@ extern "C" void foo5() {
// CHECK1-NEXT: [[TMP71:%.*]] = load ptr, ptr [[__BEGIN225]], align 8
// CHECK1-NEXT: store ptr [[TMP71]], ptr [[VV]], align 8
// CHECK1-NEXT: [[TMP72:%.*]] = load i32, ptr [[CC]], align 4
// CHECK1-NEXT: [[TMP73:%.*]] = load ptr, ptr [[VV]], align 8, !nonnull [[META5]], !align [[META6]]
// CHECK1-NEXT: [[TMP73:%.*]] = load ptr, ptr [[VV]], align 8
// CHECK1-NEXT: [[TMP74:%.*]] = load double, ptr [[TMP73]], align 8
// CHECK1-NEXT: call void (...) @body(i32 noundef [[TMP72]], double noundef [[TMP74]])
// CHECK1-NEXT: br label %[[IF_END88]]
@ -838,7 +773,7 @@ extern "C" void foo5() {
// CHECK1-NEXT: [[TMP75:%.*]] = load i64, ptr [[DOTOMP_FUSE_INDEX52]], align 8
// CHECK1-NEXT: [[INC:%.*]] = add nsw i64 [[TMP75]], 1
// CHECK1-NEXT: store i64 [[INC]], ptr [[DOTOMP_FUSE_INDEX52]], align 8
// CHECK1-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK1-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK1: [[FOR_END]]:
// CHECK1-NEXT: ret void
//
@ -903,7 +838,7 @@ extern "C" void foo5() {
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP7]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[I]], align 4
// CHECK1-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK1-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK1: [[FOR_END]]:
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_FUSE_INDEX]], align 4
// CHECK1-NEXT: br label %[[FOR_COND2:.*]]
@ -956,14 +891,14 @@ extern "C" void foo5() {
// CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_FUSE_INDEX]], align 4
// CHECK1-NEXT: [[INC16:%.*]] = add nsw i32 [[TMP24]], 1
// CHECK1-NEXT: store i32 [[INC16]], ptr [[DOTOMP_FUSE_INDEX]], align 4
// CHECK1-NEXT: br label %[[FOR_COND2]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK1-NEXT: br label %[[FOR_COND2]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK1: [[FOR_END17]]:
// CHECK1-NEXT: store i32 42, ptr [[C]], align 4
// CHECK1-NEXT: store ptr [[ARR]], ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[TMP25:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META5]], !align [[META6]]
// CHECK1-NEXT: [[TMP25:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP25]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY]], ptr [[__BEGIN2]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META5]], !align [[META6]]
// CHECK1-NEXT: [[TMP26:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY18:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP26]], i64 0, i64 0
// CHECK1-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY18]], i64 256
// CHECK1-NEXT: store ptr [[ADD_PTR]], ptr [[__END2]], align 8
@ -977,7 +912,7 @@ extern "C" void foo5() {
// CHECK1-NEXT: [[TMP29:%.*]] = load ptr, ptr [[__BEGIN2]], align 8
// CHECK1-NEXT: store ptr [[TMP29]], ptr [[V]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[C]], align 4
// CHECK1-NEXT: [[TMP31:%.*]] = load ptr, ptr [[V]], align 8, !nonnull [[META5]], !align [[META6]]
// CHECK1-NEXT: [[TMP31:%.*]] = load ptr, ptr [[V]], align 8
// CHECK1-NEXT: [[TMP32:%.*]] = load double, ptr [[TMP31]], align 8
// CHECK1-NEXT: call void (...) @body(i32 noundef [[TMP30]], double noundef [[TMP32]])
// CHECK1-NEXT: br label %[[FOR_INC22:.*]]
@ -987,8 +922,6 @@ extern "C" void foo5() {
// CHECK1-NEXT: store ptr [[INCDEC_PTR]], ptr [[__BEGIN2]], align 8
// CHECK1-NEXT: br label %[[FOR_COND19]]
// CHECK1: [[FOR_END23]]:
// CHECK1-NEXT: store i32 256, ptr [[J]], align 4
// CHECK1-NEXT: store i32 64, ptr [[K]], align 4
// CHECK1-NEXT: ret void
//
//
@ -1075,14 +1008,14 @@ extern "C" void foo5() {
// CHECK1-NEXT: store i64 [[CONV]], ptr [[DOTOMP_NI05]], align 8
// CHECK1-NEXT: store i32 42, ptr [[C]], align 4
// CHECK1-NEXT: store ptr [[ARR]], ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META5]], !align [[META6]]
// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP8]], i64 0, i64 0
// CHECK1-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 256
// CHECK1-NEXT: store ptr [[ADD_PTR]], ptr [[__END2]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META5]], !align [[META6]]
// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY7:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP9]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY7]], ptr [[__BEGIN2]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META5]], !align [[META6]]
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY9:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP10]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY9]], ptr [[DOTCAPTURE_EXPR_8]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[__END2]], align 8
@ -1132,7 +1065,7 @@ extern "C" void foo5() {
// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP22]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[I]], align 4
// CHECK1-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK1-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK1: [[FOR_END]]:
// CHECK1-NEXT: store i64 0, ptr [[DOTOMP_FUSE_INDEX29]], align 8
// CHECK1-NEXT: br label %[[FOR_COND30:.*]]
@ -1219,7 +1152,7 @@ extern "C" void foo5() {
// CHECK1-NEXT: [[TMP52:%.*]] = load ptr, ptr [[__BEGIN2]], align 8
// CHECK1-NEXT: store ptr [[TMP52]], ptr [[V]], align 8
// CHECK1-NEXT: [[TMP53:%.*]] = load i32, ptr [[C]], align 4
// CHECK1-NEXT: [[TMP54:%.*]] = load ptr, ptr [[V]], align 8, !nonnull [[META5]], !align [[META6]]
// CHECK1-NEXT: [[TMP54:%.*]] = load ptr, ptr [[V]], align 8
// CHECK1-NEXT: [[TMP55:%.*]] = load double, ptr [[TMP54]], align 8
// CHECK1-NEXT: call void (...) @body(i32 noundef [[TMP53]], double noundef [[TMP55]])
// CHECK1-NEXT: br label %[[IF_END60]]
@ -1229,14 +1162,14 @@ extern "C" void foo5() {
// CHECK1-NEXT: [[TMP56:%.*]] = load i64, ptr [[DOTOMP_FUSE_INDEX29]], align 8
// CHECK1-NEXT: [[INC62:%.*]] = add nsw i64 [[TMP56]], 1
// CHECK1-NEXT: store i64 [[INC62]], ptr [[DOTOMP_FUSE_INDEX29]], align 8
// CHECK1-NEXT: br label %[[FOR_COND30]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK1-NEXT: br label %[[FOR_COND30]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK1: [[FOR_END63]]:
// CHECK1-NEXT: store i32 37, ptr [[CC]], align 4
// CHECK1-NEXT: store ptr [[ARR]], ptr [[__RANGE264]], align 8
// CHECK1-NEXT: [[TMP57:%.*]] = load ptr, ptr [[__RANGE264]], align 8, !nonnull [[META5]], !align [[META6]]
// CHECK1-NEXT: [[TMP57:%.*]] = load ptr, ptr [[__RANGE264]], align 8
// CHECK1-NEXT: [[ARRAYDECAY66:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP57]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY66]], ptr [[__BEGIN265]], align 8
// CHECK1-NEXT: [[TMP58:%.*]] = load ptr, ptr [[__RANGE264]], align 8, !nonnull [[META5]], !align [[META6]]
// CHECK1-NEXT: [[TMP58:%.*]] = load ptr, ptr [[__RANGE264]], align 8
// CHECK1-NEXT: [[ARRAYDECAY68:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP58]], i64 0, i64 0
// CHECK1-NEXT: [[ADD_PTR69:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY68]], i64 256
// CHECK1-NEXT: store ptr [[ADD_PTR69]], ptr [[__END267]], align 8
@ -1250,7 +1183,7 @@ extern "C" void foo5() {
// CHECK1-NEXT: [[TMP61:%.*]] = load ptr, ptr [[__BEGIN265]], align 8
// CHECK1-NEXT: store ptr [[TMP61]], ptr [[VV]], align 8
// CHECK1-NEXT: [[TMP62:%.*]] = load i32, ptr [[CC]], align 4
// CHECK1-NEXT: [[TMP63:%.*]] = load ptr, ptr [[VV]], align 8, !nonnull [[META5]], !align [[META6]]
// CHECK1-NEXT: [[TMP63:%.*]] = load ptr, ptr [[VV]], align 8
// CHECK1-NEXT: [[TMP64:%.*]] = load double, ptr [[TMP63]], align 8
// CHECK1-NEXT: call void (...) @body(i32 noundef [[TMP62]], double noundef [[TMP64]])
// CHECK1-NEXT: br label %[[FOR_INC73:.*]]
@ -1260,12 +1193,6 @@ extern "C" void foo5() {
// CHECK1-NEXT: store ptr [[INCDEC_PTR]], ptr [[__BEGIN265]], align 8
// CHECK1-NEXT: br label %[[FOR_COND70]]
// CHECK1: [[FOR_END74]]:
// CHECK1-NEXT: [[TMP66:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[SUB75:%.*]] = sub nsw i32 [[TMP66]], 0
// CHECK1-NEXT: [[DIV76:%.*]] = sdiv i32 [[SUB75]], 1
// CHECK1-NEXT: [[MUL77:%.*]] = mul nsw i32 [[DIV76]], 1
// CHECK1-NEXT: [[ADD78:%.*]] = add nsw i32 0, [[MUL77]]
// CHECK1-NEXT: store i32 [[ADD78]], ptr [[DOTOMP_FUSE_INDEX]], align 4
// CHECK1-NEXT: ret void
//
//
@ -1427,34 +1354,8 @@ extern "C" void foo5() {
// CHECK2-NEXT: [[TMP43:%.*]] = load i32, ptr [[DOTOMP_FUSE_INDEX]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add i32 [[TMP43]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTOMP_FUSE_INDEX]], align 4
// CHECK2-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP2:![0-9]+]]
// CHECK2-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK2: [[FOR_END]]:
// CHECK2-NEXT: [[TMP44:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[TMP45:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK2-NEXT: [[TMP46:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[SUB28:%.*]] = sub i32 [[TMP45]], [[TMP46]]
// CHECK2-NEXT: [[SUB29:%.*]] = sub i32 [[SUB28]], 1
// CHECK2-NEXT: [[TMP47:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK2-NEXT: [[ADD30:%.*]] = add i32 [[SUB29]], [[TMP47]]
// CHECK2-NEXT: [[TMP48:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK2-NEXT: [[DIV31:%.*]] = udiv i32 [[ADD30]], [[TMP48]]
// CHECK2-NEXT: [[TMP49:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK2-NEXT: [[MUL32:%.*]] = mul i32 [[DIV31]], [[TMP49]]
// CHECK2-NEXT: [[ADD33:%.*]] = add i32 [[TMP44]], [[MUL32]]
// CHECK2-NEXT: store i32 [[ADD33]], ptr [[I]], align 4
// CHECK2-NEXT: [[TMP50:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_6]], align 4
// CHECK2-NEXT: [[TMP51:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_7]], align 4
// CHECK2-NEXT: [[TMP52:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_6]], align 4
// CHECK2-NEXT: [[SUB34:%.*]] = sub i32 [[TMP51]], [[TMP52]]
// CHECK2-NEXT: [[SUB35:%.*]] = sub i32 [[SUB34]], 1
// CHECK2-NEXT: [[TMP53:%.*]] = load i32, ptr [[DOTNEW_STEP8]], align 4
// CHECK2-NEXT: [[ADD36:%.*]] = add i32 [[SUB35]], [[TMP53]]
// CHECK2-NEXT: [[TMP54:%.*]] = load i32, ptr [[DOTNEW_STEP8]], align 4
// CHECK2-NEXT: [[DIV37:%.*]] = udiv i32 [[ADD36]], [[TMP54]]
// CHECK2-NEXT: [[TMP55:%.*]] = load i32, ptr [[DOTNEW_STEP8]], align 4
// CHECK2-NEXT: [[MUL38:%.*]] = mul i32 [[DIV37]], [[TMP55]]
// CHECK2-NEXT: [[ADD39:%.*]] = add i32 [[TMP50]], [[MUL38]]
// CHECK2-NEXT: store i32 [[ADD39]], ptr [[J]], align 4
// CHECK2-NEXT: ret void
//
//
@ -1548,14 +1449,14 @@ extern "C" void foo5() {
// CHECK2-NEXT: store i64 [[CONV]], ptr [[DOTOMP_NI05]], align 8
// CHECK2-NEXT: store i32 42, ptr [[C]], align 4
// CHECK2-NEXT: store ptr [[ARR]], ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[TMP8:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META4:![0-9]+]], !align [[META5:![0-9]+]]
// CHECK2-NEXT: [[TMP8:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP8]], i64 0, i64 0
// CHECK2-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 256
// CHECK2-NEXT: store ptr [[ADD_PTR]], ptr [[__END2]], align 8
// CHECK2-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY7:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP9]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY7]], ptr [[__BEGIN2]], align 8
// CHECK2-NEXT: [[TMP10:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP10:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY9:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP10]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY9]], ptr [[DOTCAPTURE_EXPR_8]], align 8
// CHECK2-NEXT: [[TMP11:%.*]] = load ptr, ptr [[__END2]], align 8
@ -1578,14 +1479,14 @@ extern "C" void foo5() {
// CHECK2-NEXT: store i64 [[ADD19]], ptr [[DOTOMP_NI118]], align 8
// CHECK2-NEXT: store i32 37, ptr [[CC]], align 4
// CHECK2-NEXT: store ptr [[ARR]], ptr [[__RANGE221]], align 8
// CHECK2-NEXT: [[TMP15:%.*]] = load ptr, ptr [[__RANGE221]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP15:%.*]] = load ptr, ptr [[__RANGE221]], align 8
// CHECK2-NEXT: [[ARRAYDECAY23:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP15]], i64 0, i64 0
// CHECK2-NEXT: [[ADD_PTR24:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY23]], i64 256
// CHECK2-NEXT: store ptr [[ADD_PTR24]], ptr [[__END222]], align 8
// CHECK2-NEXT: [[TMP16:%.*]] = load ptr, ptr [[__RANGE221]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP16:%.*]] = load ptr, ptr [[__RANGE221]], align 8
// CHECK2-NEXT: [[ARRAYDECAY26:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP16]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY26]], ptr [[__BEGIN225]], align 8
// CHECK2-NEXT: [[TMP17:%.*]] = load ptr, ptr [[__RANGE221]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP17:%.*]] = load ptr, ptr [[__RANGE221]], align 8
// CHECK2-NEXT: [[ARRAYDECAY28:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP17]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY28]], ptr [[DOTCAPTURE_EXPR_27]], align 8
// CHECK2-NEXT: [[TMP18:%.*]] = load ptr, ptr [[__END222]], align 8
@ -1719,7 +1620,7 @@ extern "C" void foo5() {
// CHECK2-NEXT: [[TMP60:%.*]] = load ptr, ptr [[__BEGIN2]], align 8
// CHECK2-NEXT: store ptr [[TMP60]], ptr [[V]], align 8
// CHECK2-NEXT: [[TMP61:%.*]] = load i32, ptr [[C]], align 4
// CHECK2-NEXT: [[TMP62:%.*]] = load ptr, ptr [[V]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP62:%.*]] = load ptr, ptr [[V]], align 8
// CHECK2-NEXT: [[TMP63:%.*]] = load double, ptr [[TMP62]], align 8
// CHECK2-NEXT: call void (...) @body(i32 noundef [[TMP61]], double noundef [[TMP63]])
// CHECK2-NEXT: br label %[[IF_END81]]
@ -1743,7 +1644,7 @@ extern "C" void foo5() {
// CHECK2-NEXT: [[TMP71:%.*]] = load ptr, ptr [[__BEGIN225]], align 8
// CHECK2-NEXT: store ptr [[TMP71]], ptr [[VV]], align 8
// CHECK2-NEXT: [[TMP72:%.*]] = load i32, ptr [[CC]], align 4
// CHECK2-NEXT: [[TMP73:%.*]] = load ptr, ptr [[VV]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP73:%.*]] = load ptr, ptr [[VV]], align 8
// CHECK2-NEXT: [[TMP74:%.*]] = load double, ptr [[TMP73]], align 8
// CHECK2-NEXT: call void (...) @body(i32 noundef [[TMP72]], double noundef [[TMP74]])
// CHECK2-NEXT: br label %[[IF_END88]]
@ -1753,7 +1654,7 @@ extern "C" void foo5() {
// CHECK2-NEXT: [[TMP75:%.*]] = load i64, ptr [[DOTOMP_FUSE_INDEX52]], align 8
// CHECK2-NEXT: [[INC:%.*]] = add nsw i64 [[TMP75]], 1
// CHECK2-NEXT: store i64 [[INC]], ptr [[DOTOMP_FUSE_INDEX52]], align 8
// CHECK2-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK2-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]]
// CHECK2: [[FOR_END]]:
// CHECK2-NEXT: ret void
//
@ -1818,7 +1719,7 @@ extern "C" void foo5() {
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP7]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[I]], align 4
// CHECK2-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK2-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK2: [[FOR_END]]:
// CHECK2-NEXT: store i32 0, ptr [[DOTOMP_FUSE_INDEX]], align 4
// CHECK2-NEXT: br label %[[FOR_COND2:.*]]
@ -1871,14 +1772,14 @@ extern "C" void foo5() {
// CHECK2-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_FUSE_INDEX]], align 4
// CHECK2-NEXT: [[INC16:%.*]] = add nsw i32 [[TMP24]], 1
// CHECK2-NEXT: store i32 [[INC16]], ptr [[DOTOMP_FUSE_INDEX]], align 4
// CHECK2-NEXT: br label %[[FOR_COND2]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK2-NEXT: br label %[[FOR_COND2]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK2: [[FOR_END17]]:
// CHECK2-NEXT: store i32 42, ptr [[C]], align 4
// CHECK2-NEXT: store ptr [[ARR]], ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[TMP25:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP25:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP25]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY]], ptr [[__BEGIN2]], align 8
// CHECK2-NEXT: [[TMP26:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP26:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY18:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP26]], i64 0, i64 0
// CHECK2-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY18]], i64 256
// CHECK2-NEXT: store ptr [[ADD_PTR]], ptr [[__END2]], align 8
@ -1892,7 +1793,7 @@ extern "C" void foo5() {
// CHECK2-NEXT: [[TMP29:%.*]] = load ptr, ptr [[__BEGIN2]], align 8
// CHECK2-NEXT: store ptr [[TMP29]], ptr [[V]], align 8
// CHECK2-NEXT: [[TMP30:%.*]] = load i32, ptr [[C]], align 4
// CHECK2-NEXT: [[TMP31:%.*]] = load ptr, ptr [[V]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP31:%.*]] = load ptr, ptr [[V]], align 8
// CHECK2-NEXT: [[TMP32:%.*]] = load double, ptr [[TMP31]], align 8
// CHECK2-NEXT: call void (...) @body(i32 noundef [[TMP30]], double noundef [[TMP32]])
// CHECK2-NEXT: br label %[[FOR_INC22:.*]]
@ -1902,8 +1803,6 @@ extern "C" void foo5() {
// CHECK2-NEXT: store ptr [[INCDEC_PTR]], ptr [[__BEGIN2]], align 8
// CHECK2-NEXT: br label %[[FOR_COND19]]
// CHECK2: [[FOR_END23]]:
// CHECK2-NEXT: store i32 256, ptr [[J]], align 4
// CHECK2-NEXT: store i32 64, ptr [[K]], align 4
// CHECK2-NEXT: ret void
//
//
@ -1990,14 +1889,14 @@ extern "C" void foo5() {
// CHECK2-NEXT: store i64 [[CONV]], ptr [[DOTOMP_NI05]], align 8
// CHECK2-NEXT: store i32 42, ptr [[C]], align 4
// CHECK2-NEXT: store ptr [[ARR]], ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[TMP8:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP8:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP8]], i64 0, i64 0
// CHECK2-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 256
// CHECK2-NEXT: store ptr [[ADD_PTR]], ptr [[__END2]], align 8
// CHECK2-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY7:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP9]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY7]], ptr [[__BEGIN2]], align 8
// CHECK2-NEXT: [[TMP10:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP10:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY9:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP10]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY9]], ptr [[DOTCAPTURE_EXPR_8]], align 8
// CHECK2-NEXT: [[TMP11:%.*]] = load ptr, ptr [[__END2]], align 8
@ -2047,7 +1946,7 @@ extern "C" void foo5() {
// CHECK2-NEXT: [[TMP22:%.*]] = load i32, ptr [[I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP22]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[I]], align 4
// CHECK2-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK2-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK2: [[FOR_END]]:
// CHECK2-NEXT: store i64 0, ptr [[DOTOMP_FUSE_INDEX29]], align 8
// CHECK2-NEXT: br label %[[FOR_COND30:.*]]
@ -2134,7 +2033,7 @@ extern "C" void foo5() {
// CHECK2-NEXT: [[TMP52:%.*]] = load ptr, ptr [[__BEGIN2]], align 8
// CHECK2-NEXT: store ptr [[TMP52]], ptr [[V]], align 8
// CHECK2-NEXT: [[TMP53:%.*]] = load i32, ptr [[C]], align 4
// CHECK2-NEXT: [[TMP54:%.*]] = load ptr, ptr [[V]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP54:%.*]] = load ptr, ptr [[V]], align 8
// CHECK2-NEXT: [[TMP55:%.*]] = load double, ptr [[TMP54]], align 8
// CHECK2-NEXT: call void (...) @body(i32 noundef [[TMP53]], double noundef [[TMP55]])
// CHECK2-NEXT: br label %[[IF_END60]]
@ -2144,14 +2043,14 @@ extern "C" void foo5() {
// CHECK2-NEXT: [[TMP56:%.*]] = load i64, ptr [[DOTOMP_FUSE_INDEX29]], align 8
// CHECK2-NEXT: [[INC62:%.*]] = add nsw i64 [[TMP56]], 1
// CHECK2-NEXT: store i64 [[INC62]], ptr [[DOTOMP_FUSE_INDEX29]], align 8
// CHECK2-NEXT: br label %[[FOR_COND30]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK2-NEXT: br label %[[FOR_COND30]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK2: [[FOR_END63]]:
// CHECK2-NEXT: store i32 37, ptr [[CC]], align 4
// CHECK2-NEXT: store ptr [[ARR]], ptr [[__RANGE264]], align 8
// CHECK2-NEXT: [[TMP57:%.*]] = load ptr, ptr [[__RANGE264]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP57:%.*]] = load ptr, ptr [[__RANGE264]], align 8
// CHECK2-NEXT: [[ARRAYDECAY66:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP57]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY66]], ptr [[__BEGIN265]], align 8
// CHECK2-NEXT: [[TMP58:%.*]] = load ptr, ptr [[__RANGE264]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP58:%.*]] = load ptr, ptr [[__RANGE264]], align 8
// CHECK2-NEXT: [[ARRAYDECAY68:%.*]] = getelementptr inbounds [256 x double], ptr [[TMP58]], i64 0, i64 0
// CHECK2-NEXT: [[ADD_PTR69:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY68]], i64 256
// CHECK2-NEXT: store ptr [[ADD_PTR69]], ptr [[__END267]], align 8
@ -2165,7 +2064,7 @@ extern "C" void foo5() {
// CHECK2-NEXT: [[TMP61:%.*]] = load ptr, ptr [[__BEGIN265]], align 8
// CHECK2-NEXT: store ptr [[TMP61]], ptr [[VV]], align 8
// CHECK2-NEXT: [[TMP62:%.*]] = load i32, ptr [[CC]], align 4
// CHECK2-NEXT: [[TMP63:%.*]] = load ptr, ptr [[VV]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP63:%.*]] = load ptr, ptr [[VV]], align 8
// CHECK2-NEXT: [[TMP64:%.*]] = load double, ptr [[TMP63]], align 8
// CHECK2-NEXT: call void (...) @body(i32 noundef [[TMP62]], double noundef [[TMP64]])
// CHECK2-NEXT: br label %[[FOR_INC73:.*]]
@ -2175,12 +2074,6 @@ extern "C" void foo5() {
// CHECK2-NEXT: store ptr [[INCDEC_PTR]], ptr [[__BEGIN265]], align 8
// CHECK2-NEXT: br label %[[FOR_COND70]]
// CHECK2: [[FOR_END74]]:
// CHECK2-NEXT: [[TMP66:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[SUB75:%.*]] = sub nsw i32 [[TMP66]], 0
// CHECK2-NEXT: [[DIV76:%.*]] = sdiv i32 [[SUB75]], 1
// CHECK2-NEXT: [[MUL77:%.*]] = mul nsw i32 [[DIV76]], 1
// CHECK2-NEXT: [[ADD78:%.*]] = add nsw i32 0, [[MUL77]]
// CHECK2-NEXT: store i32 [[ADD78]], ptr [[DOTOMP_FUSE_INDEX]], align 4
// CHECK2-NEXT: ret void
//
//
@ -2410,69 +2303,26 @@ extern "C" void foo5() {
// CHECK2-NEXT: [[TMP68:%.*]] = load i32, ptr [[DOTOMP_FUSE_INDEX]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add i32 [[TMP68]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTOMP_FUSE_INDEX]], align 4
// CHECK2-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK2-NEXT: br label %[[FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK2: [[FOR_END]]:
// CHECK2-NEXT: [[TMP69:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[TMP70:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK2-NEXT: [[TMP71:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[SUB53:%.*]] = sub i32 [[TMP70]], [[TMP71]]
// CHECK2-NEXT: [[SUB54:%.*]] = sub i32 [[SUB53]], 1
// CHECK2-NEXT: [[TMP72:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK2-NEXT: [[ADD55:%.*]] = add i32 [[SUB54]], [[TMP72]]
// CHECK2-NEXT: [[TMP73:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK2-NEXT: [[DIV56:%.*]] = udiv i32 [[ADD55]], [[TMP73]]
// CHECK2-NEXT: [[TMP74:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK2-NEXT: [[MUL57:%.*]] = mul i32 [[DIV56]], [[TMP74]]
// CHECK2-NEXT: [[ADD58:%.*]] = add i32 [[TMP69]], [[MUL57]]
// CHECK2-NEXT: store i32 [[ADD58]], ptr [[I]], align 4
// CHECK2-NEXT: [[TMP75:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_6]], align 4
// CHECK2-NEXT: [[TMP76:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_6]], align 4
// CHECK2-NEXT: [[TMP77:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_7]], align 4
// CHECK2-NEXT: [[SUB59:%.*]] = sub i32 [[TMP76]], [[TMP77]]
// CHECK2-NEXT: [[SUB60:%.*]] = sub i32 [[SUB59]], 1
// CHECK2-NEXT: [[TMP78:%.*]] = load i32, ptr [[DOTNEW_STEP8]], align 4
// CHECK2-NEXT: [[ADD61:%.*]] = add i32 [[SUB60]], [[TMP78]]
// CHECK2-NEXT: [[TMP79:%.*]] = load i32, ptr [[DOTNEW_STEP8]], align 4
// CHECK2-NEXT: [[DIV62:%.*]] = udiv i32 [[ADD61]], [[TMP79]]
// CHECK2-NEXT: [[TMP80:%.*]] = load i32, ptr [[DOTNEW_STEP8]], align 4
// CHECK2-NEXT: [[MUL63:%.*]] = mul i32 [[DIV62]], [[TMP80]]
// CHECK2-NEXT: [[SUB64:%.*]] = sub i32 [[TMP75]], [[MUL63]]
// CHECK2-NEXT: store i32 [[SUB64]], ptr [[J]], align 4
// CHECK2-NEXT: [[TMP81:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_17]], align 4
// CHECK2-NEXT: [[TMP82:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_19]], align 4
// CHECK2-NEXT: [[TMP83:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_17]], align 4
// CHECK2-NEXT: [[SUB65:%.*]] = sub i32 [[TMP82]], [[TMP83]]
// CHECK2-NEXT: [[SUB66:%.*]] = sub i32 [[SUB65]], 1
// CHECK2-NEXT: [[TMP84:%.*]] = load i32, ptr [[DOTNEW_STEP21]], align 4
// CHECK2-NEXT: [[ADD67:%.*]] = add i32 [[SUB66]], [[TMP84]]
// CHECK2-NEXT: [[TMP85:%.*]] = load i32, ptr [[DOTNEW_STEP21]], align 4
// CHECK2-NEXT: [[DIV68:%.*]] = udiv i32 [[ADD67]], [[TMP85]]
// CHECK2-NEXT: [[TMP86:%.*]] = load i32, ptr [[DOTNEW_STEP21]], align 4
// CHECK2-NEXT: [[MUL69:%.*]] = mul i32 [[DIV68]], [[TMP86]]
// CHECK2-NEXT: [[ADD70:%.*]] = add i32 [[TMP81]], [[MUL69]]
// CHECK2-NEXT: store i32 [[ADD70]], ptr [[K]], align 4
// CHECK2-NEXT: ret void
//
//.
// CHECK1: [[LOOP2]] = distinct !{[[LOOP2]], [[META3:![0-9]+]]}
// CHECK1: [[META3]] = !{!"llvm.loop.mustprogress"}
// CHECK1: [[LOOP4]] = distinct !{[[LOOP4]], [[META3]]}
// CHECK1: [[META5]] = !{}
// CHECK1: [[META6]] = !{i64 8}
// CHECK1: [[LOOP7]] = distinct !{[[LOOP7]], [[META3]]}
// CHECK1: [[LOOP8]] = distinct !{[[LOOP8]], [[META3]]}
// CHECK1: [[LOOP9]] = distinct !{[[LOOP9]], [[META3]]}
// CHECK1: [[LOOP10]] = distinct !{[[LOOP10]], [[META3]]}
// CHECK1: [[LOOP11]] = distinct !{[[LOOP11]], [[META3]]}
// CHECK1: [[LOOP3]] = distinct !{[[LOOP3]], [[META4:![0-9]+]]}
// CHECK1: [[META4]] = !{!"llvm.loop.mustprogress"}
// CHECK1: [[LOOP5]] = distinct !{[[LOOP5]], [[META4]]}
// CHECK1: [[LOOP6]] = distinct !{[[LOOP6]], [[META4]]}
// CHECK1: [[LOOP7]] = distinct !{[[LOOP7]], [[META4]]}
// CHECK1: [[LOOP8]] = distinct !{[[LOOP8]], [[META4]]}
// CHECK1: [[LOOP9]] = distinct !{[[LOOP9]], [[META4]]}
// CHECK1: [[LOOP10]] = distinct !{[[LOOP10]], [[META4]]}
//.
// CHECK2: [[LOOP2]] = distinct !{[[LOOP2]], [[META3:![0-9]+]]}
// CHECK2: [[META3]] = !{!"llvm.loop.mustprogress"}
// CHECK2: [[META4]] = !{}
// CHECK2: [[META5]] = !{i64 8}
// CHECK2: [[LOOP6]] = distinct !{[[LOOP6]], [[META3]]}
// CHECK2: [[LOOP7]] = distinct !{[[LOOP7]], [[META3]]}
// CHECK2: [[LOOP8]] = distinct !{[[LOOP8]], [[META3]]}
// CHECK2: [[LOOP9]] = distinct !{[[LOOP9]], [[META3]]}
// CHECK2: [[LOOP10]] = distinct !{[[LOOP10]], [[META3]]}
// CHECK2: [[LOOP11]] = distinct !{[[LOOP11]], [[META3]]}
// CHECK2: [[LOOP3]] = distinct !{[[LOOP3]], [[META4:![0-9]+]]}
// CHECK2: [[META4]] = !{!"llvm.loop.mustprogress"}
// CHECK2: [[LOOP5]] = distinct !{[[LOOP5]], [[META4]]}
// CHECK2: [[LOOP6]] = distinct !{[[LOOP6]], [[META4]]}
// CHECK2: [[LOOP7]] = distinct !{[[LOOP7]], [[META4]]}
// CHECK2: [[LOOP8]] = distinct !{[[LOOP8]], [[META4]]}
// CHECK2: [[LOOP9]] = distinct !{[[LOOP9]], [[META4]]}
// CHECK2: [[LOOP10]] = distinct !{[[LOOP10]], [[META4]]}
//.

290
clang/test/OpenMP/interchange_codegen.cpp
View file

@ -1,4 +1,4 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _ --version 4
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _
// expected-no-diagnostics
@ -123,15 +123,14 @@ extern "C" void foo10() {
#endif /* HEADER */
// CHECK1-LABEL: define dso_local void @body(
// CHECK1-SAME: ...) #[[ATTR0:[0-9]+]] {
// CHECK1-LABEL: define {{[^@]+}}@body
// CHECK1-SAME: (...) #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define dso_local void @foo1(
// CHECK1-SAME: i32 noundef [[START:%.*]], i32 noundef [[END:%.*]], i32 noundef [[STEP:%.*]]) #[[ATTR0]] {
// CHECK1-LABEL: define {{[^@]+}}@foo1
// CHECK1-SAME: (i32 noundef [[START:%.*]], i32 noundef [[END:%.*]], i32 noundef [[STEP:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[START_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[END_ADDR:%.*]] = alloca i32, align 4
@ -157,13 +156,13 @@ extern "C" void foo10() {
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP5]], [[TMP4]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP2:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define dso_local void @foo2(
// CHECK1-SAME: i32 noundef [[START1:%.*]], i32 noundef [[START2:%.*]], i32 noundef [[END1:%.*]], i32 noundef [[END2:%.*]], i32 noundef [[STEP1:%.*]], i32 noundef [[STEP2:%.*]]) #[[ATTR0]] {
// CHECK1-LABEL: define {{[^@]+}}@foo2
// CHECK1-SAME: (i32 noundef [[START1:%.*]], i32 noundef [[START2:%.*]], i32 noundef [[END1:%.*]], i32 noundef [[END2:%.*]], i32 noundef [[STEP1:%.*]], i32 noundef [[STEP2:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[START1_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[START2_ADDR:%.*]] = alloca i32, align 4
@ -263,46 +262,20 @@ extern "C" void foo10() {
// CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTPERMUTED_1_IV_I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add i32 [[TMP28]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTPERMUTED_1_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND16]], !llvm.loop [[LOOP4:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND16]], !llvm.loop [[LOOP5:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC22:%.*]]
// CHECK1: for.inc22:
// CHECK1-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTPERMUTED_0_IV_J]], align 4
// CHECK1-NEXT: [[INC23:%.*]] = add i32 [[TMP29]], 1
// CHECK1-NEXT: store i32 [[INC23]], ptr [[DOTPERMUTED_0_IV_J]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK1: for.end24:
// CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[SUB25:%.*]] = sub i32 [[TMP31]], [[TMP32]]
// CHECK1-NEXT: [[SUB26:%.*]] = sub i32 [[SUB25]], 1
// CHECK1-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK1-NEXT: [[ADD27:%.*]] = add i32 [[SUB26]], [[TMP33]]
// CHECK1-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK1-NEXT: [[DIV28:%.*]] = udiv i32 [[ADD27]], [[TMP34]]
// CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK1-NEXT: [[MUL29:%.*]] = mul i32 [[DIV28]], [[TMP35]]
// CHECK1-NEXT: [[ADD30:%.*]] = add i32 [[TMP30]], [[MUL29]]
// CHECK1-NEXT: store i32 [[ADD30]], ptr [[I]], align 4
// CHECK1-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4
// CHECK1-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_6]], align 4
// CHECK1-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4
// CHECK1-NEXT: [[SUB31:%.*]] = sub i32 [[TMP37]], [[TMP38]]
// CHECK1-NEXT: [[SUB32:%.*]] = sub i32 [[SUB31]], 1
// CHECK1-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTNEW_STEP7]], align 4
// CHECK1-NEXT: [[ADD33:%.*]] = add i32 [[SUB32]], [[TMP39]]
// CHECK1-NEXT: [[TMP40:%.*]] = load i32, ptr [[DOTNEW_STEP7]], align 4
// CHECK1-NEXT: [[DIV34:%.*]] = udiv i32 [[ADD33]], [[TMP40]]
// CHECK1-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTNEW_STEP7]], align 4
// CHECK1-NEXT: [[MUL35:%.*]] = mul i32 [[DIV34]], [[TMP41]]
// CHECK1-NEXT: [[ADD36:%.*]] = add i32 [[TMP36]], [[MUL35]]
// CHECK1-NEXT: store i32 [[ADD36]], ptr [[J]], align 4
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define dso_local void @foo3(
// CHECK1-SAME: ) #[[ATTR0]] {
// CHECK1-LABEL: define {{[^@]+}}@foo3
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
@ -369,7 +342,7 @@ extern "C" void foo10() {
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTPERMUTED_1_IV_I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTPERMUTED_1_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
@ -387,8 +360,8 @@ extern "C" void foo10() {
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define dso_local void @foo4(
// CHECK1-SAME: ) #[[ATTR0]] {
// CHECK1-LABEL: define {{[^@]+}}@foo4
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
@ -466,7 +439,7 @@ extern "C" void foo10() {
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTPERMUTED_1_IV_I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTPERMUTED_1_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
@ -484,8 +457,8 @@ extern "C" void foo10() {
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define dso_local void @foo5(
// CHECK1-SAME: ) #[[ATTR0]] {
// CHECK1-LABEL: define {{[^@]+}}@foo5
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
@ -574,8 +547,8 @@ extern "C" void foo10() {
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define dso_local void @foo6(
// CHECK1-SAME: ) #[[ATTR0]] {
// CHECK1-LABEL: define {{[^@]+}}@foo6
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
@ -712,8 +685,8 @@ extern "C" void foo10() {
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define dso_local void @foo7(
// CHECK1-SAME: ) #[[ATTR0]] {
// CHECK1-LABEL: define {{[^@]+}}@foo7
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[J:%.*]] = alloca i32, align 4
@ -781,46 +754,42 @@ extern "C" void foo10() {
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTPERMUTED_3_IV_I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTPERMUTED_3_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND11]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND11]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC16:%.*]]
// CHECK1: for.inc16:
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTPERMUTED_2_IV_L]], align 4
// CHECK1-NEXT: [[INC17:%.*]] = add nsw i32 [[TMP13]], 1
// CHECK1-NEXT: store i32 [[INC17]], ptr [[DOTPERMUTED_2_IV_L]], align 4
// CHECK1-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK1: for.end18:
// CHECK1-NEXT: br label [[FOR_INC19:%.*]]
// CHECK1: for.inc19:
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTPERMUTED_1_IV_K]], align 4
// CHECK1-NEXT: [[INC20:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK1-NEXT: store i32 [[INC20]], ptr [[DOTPERMUTED_1_IV_K]], align 4
// CHECK1-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK1: for.end21:
// CHECK1-NEXT: br label [[FOR_INC22:%.*]]
// CHECK1: for.inc22:
// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTPERMUTED_0_IV_J]], align 4
// CHECK1-NEXT: [[INC23:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK1-NEXT: store i32 [[INC23]], ptr [[DOTPERMUTED_0_IV_J]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK1: for.end24:
// CHECK1-NEXT: store i32 19, ptr [[I]], align 4
// CHECK1-NEXT: store i32 19, ptr [[J]], align 4
// CHECK1-NEXT: store i32 19, ptr [[K]], align 4
// CHECK1-NEXT: store i32 19, ptr [[L]], align 4
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define dso_local void @tfoo8(
// CHECK1-SAME: ) #[[ATTR0]] {
// CHECK1-LABEL: define {{[^@]+}}@tfoo8
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: call void @_Z4foo8ILi32EEviii(i32 noundef 0, i32 noundef 42, i32 noundef 1)
// CHECK1-NEXT: call void @_Z4foo8ILi64EEviii(i32 noundef 0, i32 noundef 42, i32 noundef 3)
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define linkonce_odr void @_Z4foo8ILi32EEviii(
// CHECK1-SAME: i32 noundef [[START:%.*]], i32 noundef [[END:%.*]], i32 noundef [[STEP:%.*]]) #[[ATTR0]] comdat {
// CHECK1-LABEL: define {{[^@]+}}@_Z4foo8ILi32EEviii
// CHECK1-SAME: (i32 noundef [[START:%.*]], i32 noundef [[END:%.*]], i32 noundef [[STEP:%.*]]) #[[ATTR0]] comdat {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[START_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[END_ADDR:%.*]] = alloca i32, align 4
@ -1529,8 +1498,8 @@ extern "C" void foo10() {
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define linkonce_odr void @_Z4foo8ILi64EEviii(
// CHECK1-SAME: i32 noundef [[START:%.*]], i32 noundef [[END:%.*]], i32 noundef [[STEP:%.*]]) #[[ATTR0]] comdat {
// CHECK1-LABEL: define {{[^@]+}}@_Z4foo8ILi64EEviii
// CHECK1-SAME: (i32 noundef [[START:%.*]], i32 noundef [[END:%.*]], i32 noundef [[STEP:%.*]]) #[[ATTR0]] comdat {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[START_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[END_ADDR:%.*]] = alloca i32, align 4
@ -2239,8 +2208,8 @@ extern "C" void foo10() {
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define dso_local void @foo9(
// CHECK1-SAME: ) #[[ATTR0]] {
// CHECK1-LABEL: define {{[^@]+}}@foo9
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[ARR:%.*]] = alloca [128 x double], align 16
// CHECK1-NEXT: [[C:%.*]] = alloca double, align 8
@ -2256,14 +2225,14 @@ extern "C" void foo10() {
// CHECK1-NEXT: [[V:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store double 4.200000e+01, ptr [[C]], align 8
// CHECK1-NEXT: store ptr [[ARR]], ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META12:![0-9]+]], !align [[META13:![0-9]+]]
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP0]], i64 0, i64 0
// CHECK1-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 128
// CHECK1-NEXT: store ptr [[ADD_PTR]], ptr [[__END2]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META12]], !align [[META13]]
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY1:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP1]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY1]], ptr [[__BEGIN2]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META12]], !align [[META13]]
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY2:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY2]], ptr [[DOTCAPTURE_EXPR_]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__END2]], align 8
@ -2308,7 +2277,7 @@ extern "C" void foo10() {
// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[__BEGIN2]], align 8
// CHECK1-NEXT: store ptr [[TMP12]], ptr [[V]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = load double, ptr [[C]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[V]], align 8, !nonnull [[META12]], !align [[META13]]
// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[V]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = load double, ptr [[TMP14]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT: call void (...) @body(double noundef [[TMP13]], double noundef [[TMP15]], i32 noundef [[TMP16]])
@ -2317,21 +2286,20 @@ extern "C" void foo10() {
// CHECK1-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTPERMUTED_1_IV___BEGIN2]], align 8
// CHECK1-NEXT: [[INC:%.*]] = add nsw i64 [[TMP17]], 1
// CHECK1-NEXT: store i64 [[INC]], ptr [[DOTPERMUTED_1_IV___BEGIN2]], align 8
// CHECK1-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP14:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP13:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC13:%.*]]
// CHECK1: for.inc13:
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTPERMUTED_0_IV_I]], align 4
// CHECK1-NEXT: [[INC14:%.*]] = add nsw i32 [[TMP18]], 1
// CHECK1-NEXT: store i32 [[INC14]], ptr [[DOTPERMUTED_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]]
// CHECK1: for.end15:
// CHECK1-NEXT: store i32 42, ptr [[I]], align 4
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define dso_local void @foo10(
// CHECK1-SAME: ) #[[ATTR0]] {
// CHECK1-LABEL: define {{[^@]+}}@foo10
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[A:%.*]] = alloca [128 x double], align 16
// CHECK1-NEXT: [[B:%.*]] = alloca [16 x double], align 16
@ -2374,14 +2342,14 @@ extern "C" void foo10() {
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2]])
// CHECK1-NEXT: store double 4.200000e+01, ptr [[C]], align 8
// CHECK1-NEXT: store ptr [[A]], ptr [[__RANGE3]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE3]], align 8, !nonnull [[META12]], !align [[META13]]
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE3]], align 8
// CHECK1-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP1]], i64 0, i64 0
// CHECK1-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 128
// CHECK1-NEXT: store ptr [[ADD_PTR]], ptr [[__END3]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE3]], align 8, !nonnull [[META12]], !align [[META13]]
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE3]], align 8
// CHECK1-NEXT: [[ARRAYDECAY4:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY4]], ptr [[__BEGIN3]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__RANGE3]], align 8, !nonnull [[META12]], !align [[META13]]
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__RANGE3]], align 8
// CHECK1-NEXT: [[ARRAYDECAY5:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY5]], ptr [[DOTCAPTURE_EXPR_]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[__END3]], align 8
@ -2399,14 +2367,14 @@ extern "C" void foo10() {
// CHECK1-NEXT: store i64 [[SUB8]], ptr [[DOTCAPTURE_EXPR_7]], align 8
// CHECK1-NEXT: store double 4.200000e+01, ptr [[D]], align 8
// CHECK1-NEXT: store ptr [[B]], ptr [[__RANGE4]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[__RANGE4]], align 8, !nonnull [[META12]], !align [[META13]]
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[__RANGE4]], align 8
// CHECK1-NEXT: [[ARRAYDECAY9:%.*]] = getelementptr inbounds [16 x double], ptr [[TMP7]], i64 0, i64 0
// CHECK1-NEXT: [[ADD_PTR10:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY9]], i64 16
// CHECK1-NEXT: store ptr [[ADD_PTR10]], ptr [[__END4]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[__RANGE4]], align 8, !nonnull [[META12]], !align [[META13]]
// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[__RANGE4]], align 8
// CHECK1-NEXT: [[ARRAYDECAY11:%.*]] = getelementptr inbounds [16 x double], ptr [[TMP8]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY11]], ptr [[__BEGIN4]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__RANGE4]], align 8, !nonnull [[META12]], !align [[META13]]
// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__RANGE4]], align 8
// CHECK1-NEXT: [[ARRAYDECAY13:%.*]] = getelementptr inbounds [16 x double], ptr [[TMP9]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY13]], ptr [[DOTCAPTURE_EXPR_12]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[__END4]], align 8
@ -2733,7 +2701,7 @@ extern "C" void foo10() {
// CHECK1-NEXT: [[TMP87:%.*]] = load double, ptr [[C]], align 8
// CHECK1-NEXT: [[TMP88:%.*]] = load double, ptr [[AA]], align 8
// CHECK1-NEXT: [[TMP89:%.*]] = load double, ptr [[D]], align 8
// CHECK1-NEXT: [[TMP90:%.*]] = load ptr, ptr [[BB]], align 8, !nonnull [[META12]], !align [[META13]]
// CHECK1-NEXT: [[TMP90:%.*]] = load ptr, ptr [[BB]], align 8
// CHECK1-NEXT: [[TMP91:%.*]] = load double, ptr [[TMP90]], align 8
// CHECK1-NEXT: [[TMP92:%.*]] = load i32, ptr [[J40]], align 4
// CHECK1-NEXT: call void (...) @body(i32 noundef [[TMP86]], double noundef [[TMP87]], double noundef [[TMP88]], double noundef [[TMP89]], double noundef [[TMP91]], i32 noundef [[TMP92]])
@ -2755,14 +2723,14 @@ extern "C" void foo10() {
// CHECK1-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @body(
// CHECK2-SAME: ...) #[[ATTR0:[0-9]+]] {
// CHECK2-LABEL: define {{[^@]+}}@body
// CHECK2-SAME: (...) #[[ATTR0:[0-9]+]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @foo1(
// CHECK2-SAME: i32 noundef [[START:%.*]], i32 noundef [[END:%.*]], i32 noundef [[STEP:%.*]]) #[[ATTR0]] {
// CHECK2-LABEL: define {{[^@]+}}@foo1
// CHECK2-SAME: (i32 noundef [[START:%.*]], i32 noundef [[END:%.*]], i32 noundef [[STEP:%.*]]) #[[ATTR0]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[START_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[END_ADDR:%.*]] = alloca i32, align 4
@ -2788,13 +2756,13 @@ extern "C" void foo10() {
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[I]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP5]], [[TMP4]]
// CHECK2-NEXT: store i32 [[ADD]], ptr [[I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP2:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @foo10(
// CHECK2-SAME: ) #[[ATTR0]] {
// CHECK2-LABEL: define {{[^@]+}}@foo10
// CHECK2-SAME: () #[[ATTR0]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[A:%.*]] = alloca [128 x double], align 16
// CHECK2-NEXT: [[B:%.*]] = alloca [16 x double], align 16
@ -2837,14 +2805,14 @@ extern "C" void foo10() {
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]])
// CHECK2-NEXT: store double 4.200000e+01, ptr [[C]], align 8
// CHECK2-NEXT: store ptr [[A]], ptr [[__RANGE3]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE3]], align 8, !nonnull [[META4:![0-9]+]], !align [[META5:![0-9]+]]
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE3]], align 8
// CHECK2-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP1]], i64 0, i64 0
// CHECK2-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 128
// CHECK2-NEXT: store ptr [[ADD_PTR]], ptr [[__END3]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE3]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE3]], align 8
// CHECK2-NEXT: [[ARRAYDECAY4:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP2]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY4]], ptr [[__BEGIN3]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__RANGE3]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__RANGE3]], align 8
// CHECK2-NEXT: [[ARRAYDECAY5:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP3]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY5]], ptr [[DOTCAPTURE_EXPR_]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[__END3]], align 8
@ -2862,14 +2830,14 @@ extern "C" void foo10() {
// CHECK2-NEXT: store i64 [[SUB8]], ptr [[DOTCAPTURE_EXPR_7]], align 8
// CHECK2-NEXT: store double 4.200000e+01, ptr [[D]], align 8
// CHECK2-NEXT: store ptr [[B]], ptr [[__RANGE4]], align 8
// CHECK2-NEXT: [[TMP7:%.*]] = load ptr, ptr [[__RANGE4]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP7:%.*]] = load ptr, ptr [[__RANGE4]], align 8
// CHECK2-NEXT: [[ARRAYDECAY9:%.*]] = getelementptr inbounds [16 x double], ptr [[TMP7]], i64 0, i64 0
// CHECK2-NEXT: [[ADD_PTR10:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY9]], i64 16
// CHECK2-NEXT: store ptr [[ADD_PTR10]], ptr [[__END4]], align 8
// CHECK2-NEXT: [[TMP8:%.*]] = load ptr, ptr [[__RANGE4]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP8:%.*]] = load ptr, ptr [[__RANGE4]], align 8
// CHECK2-NEXT: [[ARRAYDECAY11:%.*]] = getelementptr inbounds [16 x double], ptr [[TMP8]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY11]], ptr [[__BEGIN4]], align 8
// CHECK2-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__RANGE4]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__RANGE4]], align 8
// CHECK2-NEXT: [[ARRAYDECAY13:%.*]] = getelementptr inbounds [16 x double], ptr [[TMP9]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY13]], ptr [[DOTCAPTURE_EXPR_12]], align 8
// CHECK2-NEXT: [[TMP10:%.*]] = load ptr, ptr [[__END4]], align 8
@ -3196,7 +3164,7 @@ extern "C" void foo10() {
// CHECK2-NEXT: [[TMP87:%.*]] = load double, ptr [[C]], align 8
// CHECK2-NEXT: [[TMP88:%.*]] = load double, ptr [[AA]], align 8
// CHECK2-NEXT: [[TMP89:%.*]] = load double, ptr [[D]], align 8
// CHECK2-NEXT: [[TMP90:%.*]] = load ptr, ptr [[BB]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP90:%.*]] = load ptr, ptr [[BB]], align 8
// CHECK2-NEXT: [[TMP91:%.*]] = load double, ptr [[TMP90]], align 8
// CHECK2-NEXT: [[TMP92:%.*]] = load i32, ptr [[J40]], align 4
// CHECK2-NEXT: call void (...) @body(i32 noundef [[TMP86]], double noundef [[TMP87]], double noundef [[TMP88]], double noundef [[TMP89]], double noundef [[TMP91]], i32 noundef [[TMP92]])
@ -3218,8 +3186,8 @@ extern "C" void foo10() {
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @foo2(
// CHECK2-SAME: i32 noundef [[START1:%.*]], i32 noundef [[START2:%.*]], i32 noundef [[END1:%.*]], i32 noundef [[END2:%.*]], i32 noundef [[STEP1:%.*]], i32 noundef [[STEP2:%.*]]) #[[ATTR0]] {
// CHECK2-LABEL: define {{[^@]+}}@foo2
// CHECK2-SAME: (i32 noundef [[START1:%.*]], i32 noundef [[START2:%.*]], i32 noundef [[END1:%.*]], i32 noundef [[END2:%.*]], i32 noundef [[STEP1:%.*]], i32 noundef [[STEP2:%.*]]) #[[ATTR0]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[START1_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[START2_ADDR:%.*]] = alloca i32, align 4
@ -3319,46 +3287,20 @@ extern "C" void foo10() {
// CHECK2-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTPERMUTED_1_IV_I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add i32 [[TMP28]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTPERMUTED_1_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND16]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND16]], !llvm.loop [[LOOP5:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC22:%.*]]
// CHECK2: for.inc22:
// CHECK2-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTPERMUTED_0_IV_J]], align 4
// CHECK2-NEXT: [[INC23:%.*]] = add i32 [[TMP29]], 1
// CHECK2-NEXT: store i32 [[INC23]], ptr [[DOTPERMUTED_0_IV_J]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK2: for.end24:
// CHECK2-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[SUB25:%.*]] = sub i32 [[TMP31]], [[TMP32]]
// CHECK2-NEXT: [[SUB26:%.*]] = sub i32 [[SUB25]], 1
// CHECK2-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK2-NEXT: [[ADD27:%.*]] = add i32 [[SUB26]], [[TMP33]]
// CHECK2-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK2-NEXT: [[DIV28:%.*]] = udiv i32 [[ADD27]], [[TMP34]]
// CHECK2-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK2-NEXT: [[MUL29:%.*]] = mul i32 [[DIV28]], [[TMP35]]
// CHECK2-NEXT: [[ADD30:%.*]] = add i32 [[TMP30]], [[MUL29]]
// CHECK2-NEXT: store i32 [[ADD30]], ptr [[I]], align 4
// CHECK2-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4
// CHECK2-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_6]], align 4
// CHECK2-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4
// CHECK2-NEXT: [[SUB31:%.*]] = sub i32 [[TMP37]], [[TMP38]]
// CHECK2-NEXT: [[SUB32:%.*]] = sub i32 [[SUB31]], 1
// CHECK2-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTNEW_STEP7]], align 4
// CHECK2-NEXT: [[ADD33:%.*]] = add i32 [[SUB32]], [[TMP39]]
// CHECK2-NEXT: [[TMP40:%.*]] = load i32, ptr [[DOTNEW_STEP7]], align 4
// CHECK2-NEXT: [[DIV34:%.*]] = udiv i32 [[ADD33]], [[TMP40]]
// CHECK2-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTNEW_STEP7]], align 4
// CHECK2-NEXT: [[MUL35:%.*]] = mul i32 [[DIV34]], [[TMP41]]
// CHECK2-NEXT: [[ADD36:%.*]] = add i32 [[TMP36]], [[MUL35]]
// CHECK2-NEXT: store i32 [[ADD36]], ptr [[J]], align 4
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @foo3(
// CHECK2-SAME: ) #[[ATTR0]] {
// CHECK2-LABEL: define {{[^@]+}}@foo3
// CHECK2-SAME: () #[[ATTR0]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
@ -3425,7 +3367,7 @@ extern "C" void foo10() {
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTPERMUTED_1_IV_I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTPERMUTED_1_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2: omp.body.continue:
@ -3443,8 +3385,8 @@ extern "C" void foo10() {
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @foo4(
// CHECK2-SAME: ) #[[ATTR0]] {
// CHECK2-LABEL: define {{[^@]+}}@foo4
// CHECK2-SAME: () #[[ATTR0]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
@ -3522,7 +3464,7 @@ extern "C" void foo10() {
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTPERMUTED_1_IV_I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTPERMUTED_1_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2: omp.body.continue:
@ -3540,8 +3482,8 @@ extern "C" void foo10() {
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @foo5(
// CHECK2-SAME: ) #[[ATTR0]] {
// CHECK2-LABEL: define {{[^@]+}}@foo5
// CHECK2-SAME: () #[[ATTR0]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
@ -3630,8 +3572,8 @@ extern "C" void foo10() {
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @foo6(
// CHECK2-SAME: ) #[[ATTR0]] {
// CHECK2-LABEL: define {{[^@]+}}@foo6
// CHECK2-SAME: () #[[ATTR0]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
@ -3768,8 +3710,8 @@ extern "C" void foo10() {
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @foo7(
// CHECK2-SAME: ) #[[ATTR0]] {
// CHECK2-LABEL: define {{[^@]+}}@foo7
// CHECK2-SAME: () #[[ATTR0]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[J:%.*]] = alloca i32, align 4
@ -3837,38 +3779,34 @@ extern "C" void foo10() {
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTPERMUTED_3_IV_I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTPERMUTED_3_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND11]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND11]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC16:%.*]]
// CHECK2: for.inc16:
// CHECK2-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTPERMUTED_2_IV_L]], align 4
// CHECK2-NEXT: [[INC17:%.*]] = add nsw i32 [[TMP13]], 1
// CHECK2-NEXT: store i32 [[INC17]], ptr [[DOTPERMUTED_2_IV_L]], align 4
// CHECK2-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK2: for.end18:
// CHECK2-NEXT: br label [[FOR_INC19:%.*]]
// CHECK2: for.inc19:
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTPERMUTED_1_IV_K]], align 4
// CHECK2-NEXT: [[INC20:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK2-NEXT: store i32 [[INC20]], ptr [[DOTPERMUTED_1_IV_K]], align 4
// CHECK2-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK2: for.end21:
// CHECK2-NEXT: br label [[FOR_INC22:%.*]]
// CHECK2: for.inc22:
// CHECK2-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTPERMUTED_0_IV_J]], align 4
// CHECK2-NEXT: [[INC23:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK2-NEXT: store i32 [[INC23]], ptr [[DOTPERMUTED_0_IV_J]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP13:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK2: for.end24:
// CHECK2-NEXT: store i32 19, ptr [[I]], align 4
// CHECK2-NEXT: store i32 19, ptr [[J]], align 4
// CHECK2-NEXT: store i32 19, ptr [[K]], align 4
// CHECK2-NEXT: store i32 19, ptr [[L]], align 4
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @foo9(
// CHECK2-SAME: ) #[[ATTR0]] {
// CHECK2-LABEL: define {{[^@]+}}@foo9
// CHECK2-SAME: () #[[ATTR0]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[ARR:%.*]] = alloca [128 x double], align 16
// CHECK2-NEXT: [[C:%.*]] = alloca double, align 8
@ -3884,14 +3822,14 @@ extern "C" void foo10() {
// CHECK2-NEXT: [[V:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: store double 4.200000e+01, ptr [[C]], align 8
// CHECK2-NEXT: store ptr [[ARR]], ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP0]], i64 0, i64 0
// CHECK2-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 128
// CHECK2-NEXT: store ptr [[ADD_PTR]], ptr [[__END2]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY1:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP1]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY1]], ptr [[__BEGIN2]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY2:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP2]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY2]], ptr [[DOTCAPTURE_EXPR_]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__END2]], align 8
@ -3936,7 +3874,7 @@ extern "C" void foo10() {
// CHECK2-NEXT: [[TMP12:%.*]] = load ptr, ptr [[__BEGIN2]], align 8
// CHECK2-NEXT: store ptr [[TMP12]], ptr [[V]], align 8
// CHECK2-NEXT: [[TMP13:%.*]] = load double, ptr [[C]], align 8
// CHECK2-NEXT: [[TMP14:%.*]] = load ptr, ptr [[V]], align 8, !nonnull [[META4]], !align [[META5]]
// CHECK2-NEXT: [[TMP14:%.*]] = load ptr, ptr [[V]], align 8
// CHECK2-NEXT: [[TMP15:%.*]] = load double, ptr [[TMP14]], align 8
// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[I]], align 4
// CHECK2-NEXT: call void (...) @body(double noundef [[TMP13]], double noundef [[TMP15]], i32 noundef [[TMP16]])
@ -3945,29 +3883,28 @@ extern "C" void foo10() {
// CHECK2-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTPERMUTED_1_IV___BEGIN2]], align 8
// CHECK2-NEXT: [[INC:%.*]] = add nsw i64 [[TMP17]], 1
// CHECK2-NEXT: store i64 [[INC]], ptr [[DOTPERMUTED_1_IV___BEGIN2]], align 8
// CHECK2-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP14:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP13:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC13:%.*]]
// CHECK2: for.inc13:
// CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTPERMUTED_0_IV_I]], align 4
// CHECK2-NEXT: [[INC14:%.*]] = add nsw i32 [[TMP18]], 1
// CHECK2-NEXT: store i32 [[INC14]], ptr [[DOTPERMUTED_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]]
// CHECK2: for.end15:
// CHECK2-NEXT: store i32 42, ptr [[I]], align 4
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @tfoo8(
// CHECK2-SAME: ) #[[ATTR0]] {
// CHECK2-LABEL: define {{[^@]+}}@tfoo8
// CHECK2-SAME: () #[[ATTR0]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: call void @_Z4foo8ILi32EEviii(i32 noundef 0, i32 noundef 42, i32 noundef 1)
// CHECK2-NEXT: call void @_Z4foo8ILi64EEviii(i32 noundef 0, i32 noundef 42, i32 noundef 3)
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define linkonce_odr void @_Z4foo8ILi32EEviii(
// CHECK2-SAME: i32 noundef [[START:%.*]], i32 noundef [[END:%.*]], i32 noundef [[STEP:%.*]]) #[[ATTR0]] comdat {
// CHECK2-LABEL: define {{[^@]+}}@_Z4foo8ILi32EEviii
// CHECK2-SAME: (i32 noundef [[START:%.*]], i32 noundef [[END:%.*]], i32 noundef [[STEP:%.*]]) #[[ATTR0]] comdat {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[START_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[END_ADDR:%.*]] = alloca i32, align 4
@ -4676,8 +4613,8 @@ extern "C" void foo10() {
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define linkonce_odr void @_Z4foo8ILi64EEviii(
// CHECK2-SAME: i32 noundef [[START:%.*]], i32 noundef [[END:%.*]], i32 noundef [[STEP:%.*]]) #[[ATTR0]] comdat {
// CHECK2-LABEL: define {{[^@]+}}@_Z4foo8ILi64EEviii
// CHECK2-SAME: (i32 noundef [[START:%.*]], i32 noundef [[END:%.*]], i32 noundef [[STEP:%.*]]) #[[ATTR0]] comdat {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[START_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[END_ADDR:%.*]] = alloca i32, align 4
@ -5384,35 +5321,4 @@ extern "C" void foo10() {
// CHECK2: omp.precond.end:
// CHECK2-NEXT: call void @__kmpc_barrier(ptr @[[GLOB3]], i32 [[TMP0]])
// CHECK2-NEXT: ret void
//
//.
// CHECK1: [[LOOP2]] = distinct !{[[LOOP2]], [[META3:![0-9]+]]}
// CHECK1: [[META3]] = !{!"llvm.loop.mustprogress"}
// CHECK1: [[LOOP4]] = distinct !{[[LOOP4]], [[META3]]}
// CHECK1: [[LOOP5]] = distinct !{[[LOOP5]], [[META3]]}
// CHECK1: [[LOOP6]] = distinct !{[[LOOP6]], [[META3]]}
// CHECK1: [[LOOP7]] = distinct !{[[LOOP7]], [[META3]]}
// CHECK1: [[LOOP8]] = distinct !{[[LOOP8]], [[META3]]}
// CHECK1: [[LOOP9]] = distinct !{[[LOOP9]], [[META3]]}
// CHECK1: [[LOOP10]] = distinct !{[[LOOP10]], [[META3]]}
// CHECK1: [[LOOP11]] = distinct !{[[LOOP11]], [[META3]]}
// CHECK1: [[META12]] = !{}
// CHECK1: [[META13]] = !{i64 8}
// CHECK1: [[LOOP14]] = distinct !{[[LOOP14]], [[META3]]}
// CHECK1: [[LOOP15]] = distinct !{[[LOOP15]], [[META3]]}
//.
// CHECK2: [[LOOP2]] = distinct !{[[LOOP2]], [[META3:![0-9]+]]}
// CHECK2: [[META3]] = !{!"llvm.loop.mustprogress"}
// CHECK2: [[META4]] = !{}
// CHECK2: [[META5]] = !{i64 8}
// CHECK2: [[LOOP6]] = distinct !{[[LOOP6]], [[META3]]}
// CHECK2: [[LOOP7]] = distinct !{[[LOOP7]], [[META3]]}
// CHECK2: [[LOOP8]] = distinct !{[[LOOP8]], [[META3]]}
// CHECK2: [[LOOP9]] = distinct !{[[LOOP9]], [[META3]]}
// CHECK2: [[LOOP10]] = distinct !{[[LOOP10]], [[META3]]}
// CHECK2: [[LOOP11]] = distinct !{[[LOOP11]], [[META3]]}
// CHECK2: [[LOOP12]] = distinct !{[[LOOP12]], [[META3]]}
// CHECK2: [[LOOP13]] = distinct !{[[LOOP13]], [[META3]]}
// CHECK2: [[LOOP14]] = distinct !{[[LOOP14]], [[META3]]}
// CHECK2: [[LOOP15]] = distinct !{[[LOOP15]], [[META3]]}
//.

129
clang/test/OpenMP/loop_transform_final_iv.c
View file

@ -1,129 +0,0 @@
// RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=60 -emit-llvm %s -o - | FileCheck %s
// expected-no-diagnostics
// Test that loop variables are correctly finalized after loop-transformation
// constructs as required by OpenMP 6.0 spec (pg 371, lines 19-21).
void test_tile(void) {
// CHECK-LABEL: define {{.*}} @test_tile
int i;
#pragma omp tile sizes(2)
for (i = 1; i <= 10; i++) {
}
// CHECK: store i32 11, ptr %i
// After loop: i should be 11 (lb + tripcount * stride = 1 + 10 * 1)
}
void test_stripe(void) {
// CHECK-LABEL: define {{.*}} @test_stripe
int i;
#pragma omp stripe sizes(3)
for (i = 0; i < 10; i++) {
}
// CHECK: store i32 10, ptr %i
// After loop: i should be 10 (lb + tripcount * stride = 0 + 10 * 1)
}
void test_tile_nested(void) {
// CHECK-LABEL: define {{.*}} @test_tile_nested
int i, j;
#pragma omp tile sizes(2, 3)
for (i = 0; i < 10; i++)
for (j = 0; j < 5; j++) {
}
// CHECK: store i32 10, ptr %i
// CHECK: store i32 5, ptr %j
// After loop: i should be 10, j should be 5
}
void test_tile_stride(void) {
// CHECK-LABEL: define {{.*}} @test_tile_stride
int i;
#pragma omp tile sizes(4)
for (i = 5; i <= 15; i += 2) {
}
// CHECK: store i32 17, ptr %i
// After loop: i should be 17 (lb + tripcount * stride = 5 + 6 * 2)
}
void test_tile_variable_bound(int n) {
// CHECK-LABEL: define {{.*}} @test_tile_variable_bound
int i;
#pragma omp tile sizes(3)
for (i = 0; i < n; i++) {
}
// CHECK: for.end14:
// CHECK-NEXT: %{{.*}} = load i32, ptr %.capture_expr.
// CHECK-NEXT: %{{.*}} = sub nsw i32 %{{.*}}, 0
// CHECK-NEXT: %{{.*}} = sdiv i32 %{{.*}}, 1
// CHECK-NEXT: %{{.*}} = mul nsw i32 %{{.*}}, 1
// CHECK-NEXT: %{{.*}} = add nsw i32 0, %{{.*}}
// CHECK-NEXT: store i32 %{{.*}}, ptr %i
// After loop: i should equal n (the runtime upper bound)
}
void test_reverse(void) {
// CHECK-LABEL: define {{.*}} @test_reverse
int i;
#pragma omp reverse
for (i = 0; i < 10; i++) {
}
// CHECK: store i32 10, ptr %i
// After loop: i should be 10 (lb + tripcount * stride = 0 + 10 * 1)
}
void test_interchange(void) {
// CHECK-LABEL: define {{.*}} @test_interchange
int i, j;
#pragma omp interchange permutation(2, 1)
for (i = 0; i < 10; i++)
for (j = 0; j < 5; j++) {
}
// CHECK: store i32 10, ptr %i
// CHECK: store i32 5, ptr %j
// After loop: i should be 10, j should be 5
}
void test_for_workshare(void) {
// CHECK-LABEL: define {{.*}} @test_for_workshare
int i;
#pragma omp for
for (i = 5; i <= 10; i += 2) {
}
// CHECK: omp.loop.exit:
// CHECK-NEXT: call void @__kmpc_for_static_fini
// CHECK-NEXT: call void @__kmpc_barrier
// CHECK-NEXT: ret void
// omp for worksharing doesn't finalize the loop variable after the loop
}
void test_for_lastprivate(void) {
// CHECK-LABEL: define {{.*}} @test_for_lastprivate
int i;
#pragma omp parallel
#pragma omp for lastprivate(i)
for (i = 5; i <= 10; i += 2) {
}
// CHECK: .omp.lastprivate.then:
// CHECK-NEXT: store i32 11, ptr %i{{.*}}, align 4
// CHECK-NEXT: %{{.*}} = load i32, ptr %i{{.*}}, align 4
// CHECK-NEXT: store i32 %{{.*}}, ptr %{{.*}}, align 4
// omp for with lastprivate DOES finalize the loop variable
// i = last_iteration_value + stride = 9 + 2 = 11
}
void test_tile_workshare(void) {
// CHECK-LABEL: define {{.*}} @test_tile_workshare
int i;
#pragma omp for
#pragma omp tile sizes(2)
for (i = 1; i <= 10; i++) {
}
// CHECK: omp.loop.exit:
// CHECK-NOT: store {{.*}}, ptr %i
// CHECK: call void @__kmpc_for_static_fini
// i is private to the `for` construct; its post-loop value is
// unspecified here (same as plain omp-for), so no restoring store
// is expected for i after omp.loop.exit.
}

239
clang/test/OpenMP/reverse_codegen.cpp
View file

@ -1,4 +1,4 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _ --version 4
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _
// expected-no-diagnostics
@ -94,21 +94,21 @@ extern "C" void foo7() {
#endif /* HEADER */
// CHECK1-LABEL: define dso_local void @body(
// CHECK1-SAME: ...) #[[ATTR0:[0-9]+]] {
// CHECK1-LABEL: define {{[^@]+}}@body
// CHECK1-SAME: (...) #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define internal void @__cxx_global_var_init(
// CHECK1-SAME: ) #[[ATTR1:[0-9]+]] section ".text.startup" {
// CHECK1-LABEL: define {{[^@]+}}@__cxx_global_var_init
// CHECK1-SAME: () #[[ATTR1:[0-9]+]] section ".text.startup" {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: call void @_ZN1SC1Ev(ptr noundef nonnull align 4 dereferenceable(4) @s)
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define linkonce_odr void @_ZN1SC1Ev(
// CHECK1-SAME: ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR0]] comdat align 2 {
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SC1Ev
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR0]] comdat align 2 {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
@ -117,8 +117,8 @@ extern "C" void foo7() {
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define linkonce_odr void @_ZN1SC2Ev(
// CHECK1-SAME: ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR0]] comdat align 2 {
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SC2Ev
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR0]] comdat align 2 {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[I2:%.*]] = alloca ptr, align 8
@ -143,9 +143,9 @@ extern "C" void foo7() {
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTREVERSED_IV_I]], align 4
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP2]], 3
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 7, [[MUL]]
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[I2]], align 8, !nonnull [[META2:![0-9]+]], !align [[META3:![0-9]+]]
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[I2]], align 8
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP3]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[I2]], align 8, !nonnull [[META2]], !align [[META3]]
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[I2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK1-NEXT: call void (...) @body(i32 noundef [[TMP5]])
// CHECK1-NEXT: br label [[FOR_INC:%.*]]
@ -153,15 +153,13 @@ extern "C" void foo7() {
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTFORWARD_IV_I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTFORWARD_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[I2]], align 8, !nonnull [[META2]], !align [[META3]]
// CHECK1-NEXT: store i32 19, ptr [[TMP7]], align 4
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define dso_local void @foo1(
// CHECK1-SAME: i32 noundef [[START:%.*]], i32 noundef [[END:%.*]], i32 noundef [[STEP:%.*]]) #[[ATTR0]] {
// CHECK1-LABEL: define {{[^@]+}}@foo1
// CHECK1-SAME: (i32 noundef [[START:%.*]], i32 noundef [[END:%.*]], i32 noundef [[STEP:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[START_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[END_ADDR:%.*]] = alloca i32, align 4
@ -222,26 +220,13 @@ extern "C" void foo7() {
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTFORWARD_IV_I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add i32 [[TMP16]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTFORWARD_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[SUB10:%.*]] = sub i32 [[TMP18]], [[TMP19]]
// CHECK1-NEXT: [[SUB11:%.*]] = sub i32 [[SUB10]], 1
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK1-NEXT: [[ADD12:%.*]] = add i32 [[SUB11]], [[TMP20]]
// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK1-NEXT: [[DIV13:%.*]] = udiv i32 [[ADD12]], [[TMP21]]
// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK1-NEXT: [[MUL14:%.*]] = mul i32 [[DIV13]], [[TMP22]]
// CHECK1-NEXT: [[ADD15:%.*]] = add i32 [[TMP17]], [[MUL14]]
// CHECK1-NEXT: store i32 [[ADD15]], ptr [[I]], align 4
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define dso_local void @foo2(
// CHECK1-SAME: ) #[[ATTR0]] {
// CHECK1-LABEL: define {{[^@]+}}@foo2
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
@ -308,8 +293,8 @@ extern "C" void foo7() {
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define dso_local void @foo3(
// CHECK1-SAME: ) #[[ATTR0]] {
// CHECK1-LABEL: define {{[^@]+}}@foo3
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
@ -408,15 +393,15 @@ extern "C" void foo7() {
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define dso_local void @foo4(
// CHECK1-SAME: ) #[[ATTR0]] {
// CHECK1-LABEL: define {{[^@]+}}@foo4
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB2]], i32 0, ptr @foo4.omp_outlined)
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define internal void @foo4.omp_outlined(
// CHECK1-SAME: ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK1-LABEL: define {{[^@]+}}@foo4.omp_outlined
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
@ -487,15 +472,15 @@ extern "C" void foo7() {
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define dso_local void @tfoo5(
// CHECK1-SAME: ) #[[ATTR0]] {
// CHECK1-LABEL: define {{[^@]+}}@tfoo5
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: call void @_Z4foo5IiTnT_Li3EEvS0_S0_(i32 noundef 0, i32 noundef 42)
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define linkonce_odr void @_Z4foo5IiTnT_Li3EEvS0_S0_(
// CHECK1-SAME: i32 noundef [[START:%.*]], i32 noundef [[END:%.*]]) #[[ATTR0]] comdat {
// CHECK1-LABEL: define {{[^@]+}}@_Z4foo5IiTnT_Li3EEvS0_S0_
// CHECK1-SAME: (i32 noundef [[START:%.*]], i32 noundef [[END:%.*]]) #[[ATTR0]] comdat {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[START_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[END_ADDR:%.*]] = alloca i32, align 4
@ -548,23 +533,13 @@ extern "C" void foo7() {
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTFORWARD_IV_I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add i32 [[TMP12]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTFORWARD_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[SUB10:%.*]] = sub i32 [[TMP14]], [[TMP15]]
// CHECK1-NEXT: [[SUB11:%.*]] = sub i32 [[SUB10]], 1
// CHECK1-NEXT: [[ADD12:%.*]] = add i32 [[SUB11]], 3
// CHECK1-NEXT: [[DIV13:%.*]] = udiv i32 [[ADD12]], 3
// CHECK1-NEXT: [[MUL14:%.*]] = mul i32 [[DIV13]], 3
// CHECK1-NEXT: [[ADD15:%.*]] = add i32 [[TMP13]], [[MUL14]]
// CHECK1-NEXT: store i32 [[ADD15]], ptr [[I]], align 4
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define dso_local void @foo6(
// CHECK1-SAME: ) #[[ATTR0]] {
// CHECK1-LABEL: define {{[^@]+}}@foo6
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[ARR:%.*]] = alloca [128 x double], align 16
// CHECK1-NEXT: [[C:%.*]] = alloca i32, align 4
@ -579,14 +554,14 @@ extern "C" void foo7() {
// CHECK1-NEXT: [[V:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store i32 42, ptr [[C]], align 4
// CHECK1-NEXT: store ptr [[ARR]], ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META10:![0-9]+]]
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP0]], i64 0, i64 0
// CHECK1-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 128
// CHECK1-NEXT: store ptr [[ADD_PTR]], ptr [[__END2]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META10]]
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY1:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP1]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY1]], ptr [[__BEGIN2]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META10]]
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY2:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY2]], ptr [[DOTCAPTURE_EXPR_]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__END2]], align 8
@ -624,7 +599,7 @@ extern "C" void foo7() {
// CHECK1-NEXT: store ptr [[ADD_PTR10]], ptr [[__BEGIN2]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[__BEGIN2]], align 8
// CHECK1-NEXT: store ptr [[TMP12]], ptr [[V]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[V]], align 8, !nonnull [[META2]], !align [[META10]]
// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[V]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = load double, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[C]], align 4
// CHECK1-NEXT: call void (...) @body(double noundef [[TMP14]], i32 noundef [[TMP15]])
@ -633,13 +608,13 @@ extern "C" void foo7() {
// CHECK1-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTFORWARD_IV___BEGIN2]], align 8
// CHECK1-NEXT: [[INC:%.*]] = add nsw i64 [[TMP16]], 1
// CHECK1-NEXT: store i64 [[INC]], ptr [[DOTFORWARD_IV___BEGIN2]], align 8
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define dso_local void @foo7(
// CHECK1-SAME: ) #[[ATTR0]] {
// CHECK1-LABEL: define {{[^@]+}}@foo7
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[A:%.*]] = alloca [128 x double], align 16
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
@ -670,14 +645,14 @@ extern "C" void foo7() {
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2]])
// CHECK1-NEXT: store i32 42, ptr [[C]], align 4
// CHECK1-NEXT: store ptr [[A]], ptr [[__RANGE3]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE3]], align 8, !nonnull [[META2]], !align [[META10]]
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE3]], align 8
// CHECK1-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP1]], i64 0, i64 0
// CHECK1-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 128
// CHECK1-NEXT: store ptr [[ADD_PTR]], ptr [[__END3]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE3]], align 8, !nonnull [[META2]], !align [[META10]]
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE3]], align 8
// CHECK1-NEXT: [[ARRAYDECAY3:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY3]], ptr [[__BEGIN3]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__RANGE3]], align 8, !nonnull [[META2]], !align [[META10]]
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__RANGE3]], align 8
// CHECK1-NEXT: [[ARRAYDECAY4:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY4]], ptr [[DOTCAPTURE_EXPR_]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[__END3]], align 8
@ -820,7 +795,7 @@ extern "C" void foo7() {
// CHECK1-NEXT: store ptr [[TMP36]], ptr [[V]], align 8
// CHECK1-NEXT: [[TMP37:%.*]] = load i32, ptr [[K15]], align 4
// CHECK1-NEXT: [[TMP38:%.*]] = load i32, ptr [[C]], align 4
// CHECK1-NEXT: [[TMP39:%.*]] = load ptr, ptr [[V]], align 8, !nonnull [[META2]], !align [[META10]]
// CHECK1-NEXT: [[TMP39:%.*]] = load ptr, ptr [[V]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = load double, ptr [[TMP39]], align 8
// CHECK1-NEXT: [[TMP41:%.*]] = load i32, ptr [[J17]], align 4
// CHECK1-NEXT: call void (...) @body(i32 noundef [[TMP37]], i32 noundef [[TMP38]], double noundef [[TMP40]], i32 noundef [[TMP41]])
@ -842,22 +817,22 @@ extern "C" void foo7() {
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define internal void @_GLOBAL__sub_I_reverse_codegen.cpp(
// CHECK1-SAME: ) #[[ATTR1]] section ".text.startup" {
// CHECK1-LABEL: define {{[^@]+}}@_GLOBAL__sub_I_reverse_codegen.cpp
// CHECK1-SAME: () #[[ATTR1]] section ".text.startup" {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: call void @__cxx_global_var_init()
// CHECK1-NEXT: ret void
//
//
// CHECK2-LABEL: define internal void @__cxx_global_var_init(
// CHECK2-SAME: ) #[[ATTR0:[0-9]+]] section ".text.startup" {
// CHECK2-LABEL: define {{[^@]+}}@__cxx_global_var_init
// CHECK2-SAME: () #[[ATTR0:[0-9]+]] section ".text.startup" {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: call void @_ZN1SC1Ev(ptr noundef nonnull align 4 dereferenceable(4) @s)
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define linkonce_odr void @_ZN1SC1Ev(
// CHECK2-SAME: ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SC1Ev
// CHECK2-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
@ -866,8 +841,8 @@ extern "C" void foo7() {
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define linkonce_odr void @_ZN1SC2Ev(
// CHECK2-SAME: ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SC2Ev
// CHECK2-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: [[I2:%.*]] = alloca ptr, align 8
@ -892,9 +867,9 @@ extern "C" void foo7() {
// CHECK2-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTREVERSED_IV_I]], align 4
// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP2]], 3
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 7, [[MUL]]
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[I2]], align 8, !nonnull [[META2:![0-9]+]], !align [[META3:![0-9]+]]
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[I2]], align 8
// CHECK2-NEXT: store i32 [[ADD]], ptr [[TMP3]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[I2]], align 8, !nonnull [[META2]], !align [[META3]]
// CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[I2]], align 8
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK2-NEXT: call void (...) @body(i32 noundef [[TMP5]])
// CHECK2-NEXT: br label [[FOR_INC:%.*]]
@ -902,21 +877,19 @@ extern "C" void foo7() {
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTFORWARD_IV_I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTFORWARD_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: [[TMP7:%.*]] = load ptr, ptr [[I2]], align 8, !nonnull [[META2]], !align [[META3]]
// CHECK2-NEXT: store i32 19, ptr [[TMP7]], align 4
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @body(
// CHECK2-SAME: ...) #[[ATTR1]] {
// CHECK2-LABEL: define {{[^@]+}}@body
// CHECK2-SAME: (...) #[[ATTR1]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @foo1(
// CHECK2-SAME: i32 noundef [[START:%.*]], i32 noundef [[END:%.*]], i32 noundef [[STEP:%.*]]) #[[ATTR1]] {
// CHECK2-LABEL: define {{[^@]+}}@foo1
// CHECK2-SAME: (i32 noundef [[START:%.*]], i32 noundef [[END:%.*]], i32 noundef [[STEP:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[START_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[END_ADDR:%.*]] = alloca i32, align 4
@ -977,26 +950,13 @@ extern "C" void foo7() {
// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTFORWARD_IV_I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add i32 [[TMP16]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTFORWARD_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK2-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[SUB10:%.*]] = sub i32 [[TMP18]], [[TMP19]]
// CHECK2-NEXT: [[SUB11:%.*]] = sub i32 [[SUB10]], 1
// CHECK2-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK2-NEXT: [[ADD12:%.*]] = add i32 [[SUB11]], [[TMP20]]
// CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK2-NEXT: [[DIV13:%.*]] = udiv i32 [[ADD12]], [[TMP21]]
// CHECK2-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK2-NEXT: [[MUL14:%.*]] = mul i32 [[DIV13]], [[TMP22]]
// CHECK2-NEXT: [[ADD15:%.*]] = add i32 [[TMP17]], [[MUL14]]
// CHECK2-NEXT: store i32 [[ADD15]], ptr [[I]], align 4
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @foo2(
// CHECK2-SAME: ) #[[ATTR1]] {
// CHECK2-LABEL: define {{[^@]+}}@foo2
// CHECK2-SAME: () #[[ATTR1]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
@ -1063,8 +1023,8 @@ extern "C" void foo7() {
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @foo3(
// CHECK2-SAME: ) #[[ATTR1]] {
// CHECK2-LABEL: define {{[^@]+}}@foo3
// CHECK2-SAME: () #[[ATTR1]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
@ -1163,15 +1123,15 @@ extern "C" void foo7() {
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @foo4(
// CHECK2-SAME: ) #[[ATTR1]] {
// CHECK2-LABEL: define {{[^@]+}}@foo4
// CHECK2-SAME: () #[[ATTR1]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB2]], i32 0, ptr @foo4.omp_outlined)
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define internal void @foo4.omp_outlined(
// CHECK2-SAME: ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK2-LABEL: define {{[^@]+}}@foo4.omp_outlined
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
@ -1242,8 +1202,8 @@ extern "C" void foo7() {
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @foo6(
// CHECK2-SAME: ) #[[ATTR1]] {
// CHECK2-LABEL: define {{[^@]+}}@foo6
// CHECK2-SAME: () #[[ATTR1]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[ARR:%.*]] = alloca [128 x double], align 16
// CHECK2-NEXT: [[C:%.*]] = alloca i32, align 4
@ -1258,14 +1218,14 @@ extern "C" void foo7() {
// CHECK2-NEXT: [[V:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: store i32 42, ptr [[C]], align 4
// CHECK2-NEXT: store ptr [[ARR]], ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META9:![0-9]+]]
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP0]], i64 0, i64 0
// CHECK2-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 128
// CHECK2-NEXT: store ptr [[ADD_PTR]], ptr [[__END2]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META9]]
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY1:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP1]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY1]], ptr [[__BEGIN2]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META9]]
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY2:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP2]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY2]], ptr [[DOTCAPTURE_EXPR_]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__END2]], align 8
@ -1303,7 +1263,7 @@ extern "C" void foo7() {
// CHECK2-NEXT: store ptr [[ADD_PTR10]], ptr [[__BEGIN2]], align 8
// CHECK2-NEXT: [[TMP12:%.*]] = load ptr, ptr [[__BEGIN2]], align 8
// CHECK2-NEXT: store ptr [[TMP12]], ptr [[V]], align 8
// CHECK2-NEXT: [[TMP13:%.*]] = load ptr, ptr [[V]], align 8, !nonnull [[META2]], !align [[META9]]
// CHECK2-NEXT: [[TMP13:%.*]] = load ptr, ptr [[V]], align 8
// CHECK2-NEXT: [[TMP14:%.*]] = load double, ptr [[TMP13]], align 8
// CHECK2-NEXT: [[TMP15:%.*]] = load i32, ptr [[C]], align 4
// CHECK2-NEXT: call void (...) @body(double noundef [[TMP14]], i32 noundef [[TMP15]])
@ -1312,13 +1272,13 @@ extern "C" void foo7() {
// CHECK2-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTFORWARD_IV___BEGIN2]], align 8
// CHECK2-NEXT: [[INC:%.*]] = add nsw i64 [[TMP16]], 1
// CHECK2-NEXT: store i64 [[INC]], ptr [[DOTFORWARD_IV___BEGIN2]], align 8
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @foo7(
// CHECK2-SAME: ) #[[ATTR1]] {
// CHECK2-LABEL: define {{[^@]+}}@foo7
// CHECK2-SAME: () #[[ATTR1]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[A:%.*]] = alloca [128 x double], align 16
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
@ -1349,14 +1309,14 @@ extern "C" void foo7() {
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2]])
// CHECK2-NEXT: store i32 42, ptr [[C]], align 4
// CHECK2-NEXT: store ptr [[A]], ptr [[__RANGE3]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE3]], align 8, !nonnull [[META2]], !align [[META9]]
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE3]], align 8
// CHECK2-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP1]], i64 0, i64 0
// CHECK2-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 128
// CHECK2-NEXT: store ptr [[ADD_PTR]], ptr [[__END3]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE3]], align 8, !nonnull [[META2]], !align [[META9]]
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE3]], align 8
// CHECK2-NEXT: [[ARRAYDECAY3:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP2]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY3]], ptr [[__BEGIN3]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__RANGE3]], align 8, !nonnull [[META2]], !align [[META9]]
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__RANGE3]], align 8
// CHECK2-NEXT: [[ARRAYDECAY4:%.*]] = getelementptr inbounds [128 x double], ptr [[TMP3]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY4]], ptr [[DOTCAPTURE_EXPR_]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[__END3]], align 8
@ -1499,7 +1459,7 @@ extern "C" void foo7() {
// CHECK2-NEXT: store ptr [[TMP36]], ptr [[V]], align 8
// CHECK2-NEXT: [[TMP37:%.*]] = load i32, ptr [[K15]], align 4
// CHECK2-NEXT: [[TMP38:%.*]] = load i32, ptr [[C]], align 4
// CHECK2-NEXT: [[TMP39:%.*]] = load ptr, ptr [[V]], align 8, !nonnull [[META2]], !align [[META9]]
// CHECK2-NEXT: [[TMP39:%.*]] = load ptr, ptr [[V]], align 8
// CHECK2-NEXT: [[TMP40:%.*]] = load double, ptr [[TMP39]], align 8
// CHECK2-NEXT: [[TMP41:%.*]] = load i32, ptr [[J17]], align 4
// CHECK2-NEXT: call void (...) @body(i32 noundef [[TMP37]], i32 noundef [[TMP38]], double noundef [[TMP40]], i32 noundef [[TMP41]])
@ -1521,15 +1481,15 @@ extern "C" void foo7() {
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define dso_local void @tfoo5(
// CHECK2-SAME: ) #[[ATTR1]] {
// CHECK2-LABEL: define {{[^@]+}}@tfoo5
// CHECK2-SAME: () #[[ATTR1]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: call void @_Z4foo5IiTnT_Li3EEvS0_S0_(i32 noundef 0, i32 noundef 42)
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define linkonce_odr void @_Z4foo5IiTnT_Li3EEvS0_S0_(
// CHECK2-SAME: i32 noundef [[START:%.*]], i32 noundef [[END:%.*]]) #[[ATTR1]] comdat {
// CHECK2-LABEL: define {{[^@]+}}@_Z4foo5IiTnT_Li3EEvS0_S0_
// CHECK2-SAME: (i32 noundef [[START:%.*]], i32 noundef [[END:%.*]]) #[[ATTR1]] comdat {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[START_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[END_ADDR:%.*]] = alloca i32, align 4
@ -1582,43 +1542,14 @@ extern "C" void foo7() {
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTFORWARD_IV_I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add i32 [[TMP12]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTFORWARD_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK2-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[SUB10:%.*]] = sub i32 [[TMP14]], [[TMP15]]
// CHECK2-NEXT: [[SUB11:%.*]] = sub i32 [[SUB10]], 1
// CHECK2-NEXT: [[ADD12:%.*]] = add i32 [[SUB11]], 3
// CHECK2-NEXT: [[DIV13:%.*]] = udiv i32 [[ADD12]], 3
// CHECK2-NEXT: [[MUL14:%.*]] = mul i32 [[DIV13]], 3
// CHECK2-NEXT: [[ADD15:%.*]] = add i32 [[TMP13]], [[MUL14]]
// CHECK2-NEXT: store i32 [[ADD15]], ptr [[I]], align 4
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define internal void @_GLOBAL__sub_I_reverse_codegen.cpp(
// CHECK2-SAME: ) #[[ATTR0]] section ".text.startup" {
// CHECK2-LABEL: define {{[^@]+}}@_GLOBAL__sub_I_reverse_codegen.cpp
// CHECK2-SAME: () #[[ATTR0]] section ".text.startup" {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: call void @__cxx_global_var_init()
// CHECK2-NEXT: ret void
//
//.
// CHECK1: [[META2]] = !{}
// CHECK1: [[META3]] = !{i64 4}
// CHECK1: [[LOOP4]] = distinct !{[[LOOP4]], [[META5:![0-9]+]]}
// CHECK1: [[META5]] = !{!"llvm.loop.mustprogress"}
// CHECK1: [[LOOP6]] = distinct !{[[LOOP6]], [[META5]]}
// CHECK1: [[LOOP9]] = distinct !{[[LOOP9]], [[META5]]}
// CHECK1: [[META10]] = !{i64 8}
// CHECK1: [[LOOP11]] = distinct !{[[LOOP11]], [[META5]]}
//.
// CHECK2: [[META2]] = !{}
// CHECK2: [[META3]] = !{i64 4}
// CHECK2: [[LOOP4]] = distinct !{[[LOOP4]], [[META5:![0-9]+]]}
// CHECK2: [[META5]] = !{!"llvm.loop.mustprogress"}
// CHECK2: [[LOOP6]] = distinct !{[[LOOP6]], [[META5]]}
// CHECK2: [[META9]] = !{i64 8}
// CHECK2: [[LOOP10]] = distinct !{[[LOOP10]], [[META5]]}
// CHECK2: [[LOOP11]] = distinct !{[[LOOP11]], [[META5]]}
//.

282
clang/test/OpenMP/stripe_codegen.cpp
View file

@ -179,9 +179,9 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP5]], 3
// CHECK1-NEXT: [[ADD9:%.*]] = add nsw i32 7, [[MUL]]
// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[I2]], align 8, !nonnull [[META2:![0-9]+]], !align [[META3:![0-9]+]]
// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[I2]], align 8
// CHECK1-NEXT: store i32 [[ADD9]], ptr [[TMP6]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[I2]], align 8, !nonnull [[META2]], !align [[META3]]
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[I2]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: call void (...) @body(i32 noundef [[TMP8]])
// CHECK1-NEXT: br label [[FOR_INC:%.*]]
@ -189,17 +189,15 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP9]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND4]], !llvm.loop [[LOOP4:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND4]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC10:%.*]]
// CHECK1: for.inc10:
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: [[ADD11:%.*]] = add nsw i32 [[TMP10]], 5
// CHECK1-NEXT: store i32 [[ADD11]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]]
// CHECK1: for.end12:
// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[I2]], align 8, !nonnull [[META2]], !align [[META3]]
// CHECK1-NEXT: store i32 19, ptr [[TMP11]], align 4
// CHECK1-NEXT: ret void
//
//
@ -283,28 +281,15 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add i32 [[TMP20]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC15:%.*]]
// CHECK1: for.inc15:
// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: [[ADD16:%.*]] = add i32 [[TMP21]], 5
// CHECK1-NEXT: store i32 [[ADD16]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK1: for.end17:
// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[SUB18:%.*]] = sub i32 [[TMP23]], [[TMP24]]
// CHECK1-NEXT: [[SUB19:%.*]] = sub i32 [[SUB18]], 1
// CHECK1-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK1-NEXT: [[ADD20:%.*]] = add i32 [[SUB19]], [[TMP25]]
// CHECK1-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK1-NEXT: [[DIV21:%.*]] = udiv i32 [[ADD20]], [[TMP26]]
// CHECK1-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK1-NEXT: [[MUL22:%.*]] = mul i32 [[DIV21]], [[TMP27]]
// CHECK1-NEXT: [[ADD23:%.*]] = add i32 [[TMP22]], [[MUL22]]
// CHECK1-NEXT: store i32 [[ADD23]], ptr [[I]], align 4
// CHECK1-NEXT: ret void
//
//
@ -395,31 +380,29 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTSTRIPE_1_IV_J]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTSTRIPE_1_IV_J]], align 4
// CHECK1-NEXT: br label [[FOR_COND10]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND10]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC22:%.*]]
// CHECK1: for.inc22:
// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK1-NEXT: [[INC23:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK1-NEXT: store i32 [[INC23]], ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND4]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND4]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK1: for.end24:
// CHECK1-NEXT: br label [[FOR_INC25:%.*]]
// CHECK1: for.inc25:
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK1-NEXT: [[ADD26:%.*]] = add nsw i32 [[TMP16]], 5
// CHECK1-NEXT: store i32 [[ADD26]], ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK1-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK1: for.end27:
// CHECK1-NEXT: br label [[FOR_INC28:%.*]]
// CHECK1: for.inc28:
// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP17]], 5
// CHECK1-NEXT: store i32 [[ADD29]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK1: for.end30:
// CHECK1-NEXT: store i32 19, ptr [[I]], align 4
// CHECK1-NEXT: store i32 19, ptr [[J]], align 4
// CHECK1-NEXT: ret void
//
//
@ -533,21 +516,21 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTSTRIPE_1_IV_J]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP20]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTSTRIPE_1_IV_J]], align 4
// CHECK1-NEXT: br label [[FOR_COND15]], !llvm.loop [[LOOP13:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND15]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC27:%.*]]
// CHECK1: for.inc27:
// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK1-NEXT: [[INC28:%.*]] = add nsw i32 [[TMP21]], 1
// CHECK1-NEXT: store i32 [[INC28]], ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND3]], !llvm.loop [[LOOP14:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND3]], !llvm.loop [[LOOP13:![0-9]+]]
// CHECK1: for.end29:
// CHECK1-NEXT: br label [[FOR_INC30:%.*]]
// CHECK1: for.inc30:
// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK1-NEXT: [[ADD31:%.*]] = add nsw i32 [[TMP22]], 5
// CHECK1-NEXT: store i32 [[ADD31]], ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]]
// CHECK1: for.end32:
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
@ -686,21 +669,21 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTSTRIPE_1_IV_J]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP22]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTSTRIPE_1_IV_J]], align 4
// CHECK1-NEXT: br label [[FOR_COND20]], !llvm.loop [[LOOP16:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND20]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC32:%.*]]
// CHECK1: for.inc32:
// CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK1-NEXT: [[INC33:%.*]] = add nsw i32 [[TMP23]], 1
// CHECK1-NEXT: store i32 [[INC33]], ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND8]], !llvm.loop [[LOOP17:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND8]], !llvm.loop [[LOOP16:![0-9]+]]
// CHECK1: for.end34:
// CHECK1-NEXT: br label [[FOR_INC35:%.*]]
// CHECK1: for.inc35:
// CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK1-NEXT: [[ADD36:%.*]] = add nsw i32 [[TMP24]], 5
// CHECK1-NEXT: store i32 [[ADD36]], ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP18:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP17:![0-9]+]]
// CHECK1: for.end37:
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
@ -1013,7 +996,7 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP18:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
@ -1109,25 +1092,15 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add i32 [[TMP16]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP22:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP21:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC15:%.*]]
// CHECK1: for.inc15:
// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: [[ADD16:%.*]] = add i32 [[TMP17]], 5
// CHECK1-NEXT: store i32 [[ADD16]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP23:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP22:![0-9]+]]
// CHECK1: for.end17:
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[SUB18:%.*]] = sub i32 [[TMP19]], [[TMP20]]
// CHECK1-NEXT: [[SUB19:%.*]] = sub i32 [[SUB18]], 1
// CHECK1-NEXT: [[ADD20:%.*]] = add i32 [[SUB19]], 3
// CHECK1-NEXT: [[DIV21:%.*]] = udiv i32 [[ADD20]], 3
// CHECK1-NEXT: [[MUL22:%.*]] = mul i32 [[DIV21]], 3
// CHECK1-NEXT: [[ADD23:%.*]] = add i32 [[TMP18]], [[MUL22]]
// CHECK1-NEXT: store i32 [[ADD23]], ptr [[I]], align 4
// CHECK1-NEXT: ret void
//
//
@ -1198,7 +1171,7 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP11]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP24:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP23:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC17:%.*]]
// CHECK1: for.inc17:
@ -1215,9 +1188,8 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: [[ADD23:%.*]] = add nsw i32 [[TMP14]], [[COND22]]
// CHECK1-NEXT: store i32 [[ADD23]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP25:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP24:![0-9]+]]
// CHECK1: for.end24:
// CHECK1-NEXT: store i32 19, ptr [[I]], align 4
// CHECK1-NEXT: ret void
//
//
@ -1235,14 +1207,14 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[V:%.*]] = alloca double, align 8
// CHECK1-NEXT: [[ARRAY:%.*]] = getelementptr inbounds nuw [[STRUCT_DATA_T]], ptr [[DATA]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[ARRAY]], ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META26:![0-9]+]]
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP0]], i64 0, i64 0
// CHECK1-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 12
// CHECK1-NEXT: store ptr [[ADD_PTR]], ptr [[__END2]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META26]]
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY1:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP1]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY1]], ptr [[__BEGIN2]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META26]]
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY2:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY2]], ptr [[DOTCAPTURE_EXPR_]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__END2]], align 8
@ -1306,14 +1278,14 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTSTRIPE_0_IV___BEGIN2]], align 8
// CHECK1-NEXT: [[INC:%.*]] = add nsw i64 [[TMP19]], 1
// CHECK1-NEXT: store i64 [[INC]], ptr [[DOTSTRIPE_0_IV___BEGIN2]], align 8
// CHECK1-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP27:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP25:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC16:%.*]]
// CHECK1: for.inc16:
// CHECK1-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTFLOOR_0_IV___BEGIN2]], align 8
// CHECK1-NEXT: [[ADD17:%.*]] = add nsw i64 [[TMP20]], 5
// CHECK1-NEXT: store i64 [[ADD17]], ptr [[DOTFLOOR_0_IV___BEGIN2]], align 8
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP28:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP26:![0-9]+]]
// CHECK1: for.end18:
// CHECK1-NEXT: ret void
//
@ -1334,14 +1306,14 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: store double 4.200000e+01, ptr [[C]], align 8
// CHECK1-NEXT: [[ARRAY:%.*]] = getelementptr inbounds nuw [[STRUCT_DATA_T]], ptr [[DATA]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[ARRAY]], ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META26]]
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP0]], i64 0, i64 0
// CHECK1-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 12
// CHECK1-NEXT: store ptr [[ADD_PTR]], ptr [[__END2]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META26]]
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY1:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP1]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY1]], ptr [[__BEGIN2]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META26]]
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY2:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY2]], ptr [[DOTCAPTURE_EXPR_]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__END2]], align 8
@ -1406,14 +1378,14 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTSTRIPE_0_IV___BEGIN2]], align 8
// CHECK1-NEXT: [[INC:%.*]] = add nsw i64 [[TMP20]], 1
// CHECK1-NEXT: store i64 [[INC]], ptr [[DOTSTRIPE_0_IV___BEGIN2]], align 8
// CHECK1-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP29:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP27:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC16:%.*]]
// CHECK1: for.inc16:
// CHECK1-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTFLOOR_0_IV___BEGIN2]], align 8
// CHECK1-NEXT: [[ADD17:%.*]] = add nsw i64 [[TMP21]], 5
// CHECK1-NEXT: store i64 [[ADD17]], ptr [[DOTFLOOR_0_IV___BEGIN2]], align 8
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP30:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP28:![0-9]+]]
// CHECK1: for.end18:
// CHECK1-NEXT: ret void
//
@ -1485,9 +1457,9 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP5]], 3
// CHECK2-NEXT: [[ADD9:%.*]] = add nsw i32 7, [[MUL]]
// CHECK2-NEXT: [[TMP6:%.*]] = load ptr, ptr [[I2]], align 8, !nonnull [[META2:![0-9]+]], !align [[META3:![0-9]+]]
// CHECK2-NEXT: [[TMP6:%.*]] = load ptr, ptr [[I2]], align 8
// CHECK2-NEXT: store i32 [[ADD9]], ptr [[TMP6]], align 4
// CHECK2-NEXT: [[TMP7:%.*]] = load ptr, ptr [[I2]], align 8, !nonnull [[META2]], !align [[META3]]
// CHECK2-NEXT: [[TMP7:%.*]] = load ptr, ptr [[I2]], align 8
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK2-NEXT: call void (...) @body(i32 noundef [[TMP8]])
// CHECK2-NEXT: br label [[FOR_INC:%.*]]
@ -1495,17 +1467,15 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP9]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND4]], !llvm.loop [[LOOP4:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND4]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC10:%.*]]
// CHECK2: for.inc10:
// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: [[ADD11:%.*]] = add nsw i32 [[TMP10]], 5
// CHECK2-NEXT: store i32 [[ADD11]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]]
// CHECK2: for.end12:
// CHECK2-NEXT: [[TMP11:%.*]] = load ptr, ptr [[I2]], align 8, !nonnull [[META2]], !align [[META3]]
// CHECK2-NEXT: store i32 19, ptr [[TMP11]], align 4
// CHECK2-NEXT: ret void
//
//
@ -1595,28 +1565,15 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add i32 [[TMP20]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC15:%.*]]
// CHECK2: for.inc15:
// CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: [[ADD16:%.*]] = add i32 [[TMP21]], 5
// CHECK2-NEXT: store i32 [[ADD16]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK2: for.end17:
// CHECK2-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK2-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[SUB18:%.*]] = sub i32 [[TMP23]], [[TMP24]]
// CHECK2-NEXT: [[SUB19:%.*]] = sub i32 [[SUB18]], 1
// CHECK2-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK2-NEXT: [[ADD20:%.*]] = add i32 [[SUB19]], [[TMP25]]
// CHECK2-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK2-NEXT: [[DIV21:%.*]] = udiv i32 [[ADD20]], [[TMP26]]
// CHECK2-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK2-NEXT: [[MUL22:%.*]] = mul i32 [[DIV21]], [[TMP27]]
// CHECK2-NEXT: [[ADD23:%.*]] = add i32 [[TMP22]], [[MUL22]]
// CHECK2-NEXT: store i32 [[ADD23]], ptr [[I]], align 4
// CHECK2-NEXT: ret void
//
//
@ -1636,14 +1593,14 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: store double 4.200000e+01, ptr [[C]], align 8
// CHECK2-NEXT: [[ARRAY:%.*]] = getelementptr inbounds nuw [[STRUCT_DATA_T]], ptr [[DATA]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[ARRAY]], ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META9:![0-9]+]]
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP0]], i64 0, i64 0
// CHECK2-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 12
// CHECK2-NEXT: store ptr [[ADD_PTR]], ptr [[__END2]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META9]]
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY1:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP1]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY1]], ptr [[__BEGIN2]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META9]]
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY2:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP2]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY2]], ptr [[DOTCAPTURE_EXPR_]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__END2]], align 8
@ -1708,14 +1665,14 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTSTRIPE_0_IV___BEGIN2]], align 8
// CHECK2-NEXT: [[INC:%.*]] = add nsw i64 [[TMP20]], 1
// CHECK2-NEXT: store i64 [[INC]], ptr [[DOTSTRIPE_0_IV___BEGIN2]], align 8
// CHECK2-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC16:%.*]]
// CHECK2: for.inc16:
// CHECK2-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTFLOOR_0_IV___BEGIN2]], align 8
// CHECK2-NEXT: [[ADD17:%.*]] = add nsw i64 [[TMP21]], 5
// CHECK2-NEXT: store i64 [[ADD17]], ptr [[DOTFLOOR_0_IV___BEGIN2]], align 8
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK2: for.end18:
// CHECK2-NEXT: ret void
//
@ -1807,31 +1764,29 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTSTRIPE_1_IV_J]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTSTRIPE_1_IV_J]], align 4
// CHECK2-NEXT: br label [[FOR_COND10]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND10]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC22:%.*]]
// CHECK2: for.inc22:
// CHECK2-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK2-NEXT: [[INC23:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK2-NEXT: store i32 [[INC23]], ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND4]], !llvm.loop [[LOOP13:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND4]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK2: for.end24:
// CHECK2-NEXT: br label [[FOR_INC25:%.*]]
// CHECK2: for.inc25:
// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK2-NEXT: [[ADD26:%.*]] = add nsw i32 [[TMP16]], 5
// CHECK2-NEXT: store i32 [[ADD26]], ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK2-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP14:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK2: for.end27:
// CHECK2-NEXT: br label [[FOR_INC28:%.*]]
// CHECK2: for.inc28:
// CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP17]], 5
// CHECK2-NEXT: store i32 [[ADD29]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP13:![0-9]+]]
// CHECK2: for.end30:
// CHECK2-NEXT: store i32 19, ptr [[I]], align 4
// CHECK2-NEXT: store i32 19, ptr [[J]], align 4
// CHECK2-NEXT: ret void
//
//
@ -1945,21 +1900,21 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTSTRIPE_1_IV_J]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP20]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTSTRIPE_1_IV_J]], align 4
// CHECK2-NEXT: br label [[FOR_COND15]], !llvm.loop [[LOOP16:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND15]], !llvm.loop [[LOOP14:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC27:%.*]]
// CHECK2: for.inc27:
// CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK2-NEXT: [[INC28:%.*]] = add nsw i32 [[TMP21]], 1
// CHECK2-NEXT: store i32 [[INC28]], ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND3]], !llvm.loop [[LOOP17:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND3]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK2: for.end29:
// CHECK2-NEXT: br label [[FOR_INC30:%.*]]
// CHECK2: for.inc30:
// CHECK2-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK2-NEXT: [[ADD31:%.*]] = add nsw i32 [[TMP22]], 5
// CHECK2-NEXT: store i32 [[ADD31]], ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP18:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]]
// CHECK2: for.end32:
// CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2: omp.body.continue:
@ -2098,21 +2053,21 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTSTRIPE_1_IV_J]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP22]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTSTRIPE_1_IV_J]], align 4
// CHECK2-NEXT: br label [[FOR_COND20]], !llvm.loop [[LOOP19:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND20]], !llvm.loop [[LOOP17:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC32:%.*]]
// CHECK2: for.inc32:
// CHECK2-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK2-NEXT: [[INC33:%.*]] = add nsw i32 [[TMP23]], 1
// CHECK2-NEXT: store i32 [[INC33]], ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND8]], !llvm.loop [[LOOP20:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND8]], !llvm.loop [[LOOP18:![0-9]+]]
// CHECK2: for.end34:
// CHECK2-NEXT: br label [[FOR_INC35:%.*]]
// CHECK2: for.inc35:
// CHECK2-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK2-NEXT: [[ADD36:%.*]] = add nsw i32 [[TMP24]], 5
// CHECK2-NEXT: store i32 [[ADD36]], ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP21:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
// CHECK2: for.end37:
// CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2: omp.body.continue:
@ -2425,7 +2380,7 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP22:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP20:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2: omp.body.continue:
@ -2509,7 +2464,7 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP11]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP25:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP23:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC17:%.*]]
// CHECK2: for.inc17:
@ -2526,9 +2481,8 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: [[ADD23:%.*]] = add nsw i32 [[TMP14]], [[COND22]]
// CHECK2-NEXT: store i32 [[ADD23]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP26:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP24:![0-9]+]]
// CHECK2: for.end24:
// CHECK2-NEXT: store i32 19, ptr [[I]], align 4
// CHECK2-NEXT: ret void
//
//
@ -2546,14 +2500,14 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[V:%.*]] = alloca double, align 8
// CHECK2-NEXT: [[ARRAY:%.*]] = getelementptr inbounds nuw [[STRUCT_DATA_T]], ptr [[DATA]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[ARRAY]], ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META9]]
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP0]], i64 0, i64 0
// CHECK2-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 12
// CHECK2-NEXT: store ptr [[ADD_PTR]], ptr [[__END2]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META9]]
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY1:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP1]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY1]], ptr [[__BEGIN2]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META9]]
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY2:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP2]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY2]], ptr [[DOTCAPTURE_EXPR_]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__END2]], align 8
@ -2617,14 +2571,14 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTSTRIPE_0_IV___BEGIN2]], align 8
// CHECK2-NEXT: [[INC:%.*]] = add nsw i64 [[TMP19]], 1
// CHECK2-NEXT: store i64 [[INC]], ptr [[DOTSTRIPE_0_IV___BEGIN2]], align 8
// CHECK2-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP27:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP25:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC16:%.*]]
// CHECK2: for.inc16:
// CHECK2-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTFLOOR_0_IV___BEGIN2]], align 8
// CHECK2-NEXT: [[ADD17:%.*]] = add nsw i64 [[TMP20]], 5
// CHECK2-NEXT: store i64 [[ADD17]], ptr [[DOTFLOOR_0_IV___BEGIN2]], align 8
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP28:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP26:![0-9]+]]
// CHECK2: for.end18:
// CHECK2-NEXT: ret void
//
@ -2708,25 +2662,15 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add i32 [[TMP16]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTSTRIPE_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP29:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP27:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC15:%.*]]
// CHECK2: for.inc15:
// CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: [[ADD16:%.*]] = add i32 [[TMP17]], 5
// CHECK2-NEXT: store i32 [[ADD16]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP30:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP28:![0-9]+]]
// CHECK2: for.end17:
// CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK2-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[SUB18:%.*]] = sub i32 [[TMP19]], [[TMP20]]
// CHECK2-NEXT: [[SUB19:%.*]] = sub i32 [[SUB18]], 1
// CHECK2-NEXT: [[ADD20:%.*]] = add i32 [[SUB19]], 3
// CHECK2-NEXT: [[DIV21:%.*]] = udiv i32 [[ADD20]], 3
// CHECK2-NEXT: [[MUL22:%.*]] = mul i32 [[DIV21]], 3
// CHECK2-NEXT: [[ADD23:%.*]] = add i32 [[TMP18]], [[MUL22]]
// CHECK2-NEXT: store i32 [[ADD23]], ptr [[I]], align 4
// CHECK2-NEXT: ret void
//
//
@ -2737,59 +2681,53 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: ret void
//
//.
// CHECK1: [[META2]] = !{}
// CHECK1: [[META3]] = !{i64 4}
// CHECK1: [[LOOP4]] = distinct !{[[LOOP4]], [[META5:![0-9]+]]}
// CHECK1: [[META5]] = !{!"llvm.loop.mustprogress"}
// CHECK1: [[LOOP6]] = distinct !{[[LOOP6]], [[META5]]}
// CHECK1: [[LOOP7]] = distinct !{[[LOOP7]], [[META5]]}
// CHECK1: [[LOOP8]] = distinct !{[[LOOP8]], [[META5]]}
// CHECK1: [[LOOP9]] = distinct !{[[LOOP9]], [[META5]]}
// CHECK1: [[LOOP10]] = distinct !{[[LOOP10]], [[META5]]}
// CHECK1: [[LOOP11]] = distinct !{[[LOOP11]], [[META5]]}
// CHECK1: [[LOOP12]] = distinct !{[[LOOP12]], [[META5]]}
// CHECK1: [[LOOP13]] = distinct !{[[LOOP13]], [[META5]]}
// CHECK1: [[LOOP14]] = distinct !{[[LOOP14]], [[META5]]}
// CHECK1: [[LOOP15]] = distinct !{[[LOOP15]], [[META5]]}
// CHECK1: [[LOOP16]] = distinct !{[[LOOP16]], [[META5]]}
// CHECK1: [[LOOP17]] = distinct !{[[LOOP17]], [[META5]]}
// CHECK1: [[LOOP18]] = distinct !{[[LOOP18]], [[META5]]}
// CHECK1: [[LOOP19]] = distinct !{[[LOOP19]], [[META5]]}
// CHECK1: [[LOOP22]] = distinct !{[[LOOP22]], [[META5]]}
// CHECK1: [[LOOP23]] = distinct !{[[LOOP23]], [[META5]]}
// CHECK1: [[LOOP24]] = distinct !{[[LOOP24]], [[META5]]}
// CHECK1: [[LOOP25]] = distinct !{[[LOOP25]], [[META5]]}
// CHECK1: [[META26]] = !{i64 8}
// CHECK1: [[LOOP27]] = distinct !{[[LOOP27]], [[META5]]}
// CHECK1: [[LOOP28]] = distinct !{[[LOOP28]], [[META5]]}
// CHECK1: [[LOOP29]] = distinct !{[[LOOP29]], [[META5]]}
// CHECK1: [[LOOP30]] = distinct !{[[LOOP30]], [[META5]]}
// CHECK1: [[LOOP3]] = distinct !{[[LOOP3]], [[META4:![0-9]+]]}
// CHECK1: [[META4]] = !{!"llvm.loop.mustprogress"}
// CHECK1: [[LOOP5]] = distinct !{[[LOOP5]], [[META4]]}
// CHECK1: [[LOOP6]] = distinct !{[[LOOP6]], [[META4]]}
// CHECK1: [[LOOP7]] = distinct !{[[LOOP7]], [[META4]]}
// CHECK1: [[LOOP8]] = distinct !{[[LOOP8]], [[META4]]}
// CHECK1: [[LOOP9]] = distinct !{[[LOOP9]], [[META4]]}
// CHECK1: [[LOOP10]] = distinct !{[[LOOP10]], [[META4]]}
// CHECK1: [[LOOP11]] = distinct !{[[LOOP11]], [[META4]]}
// CHECK1: [[LOOP12]] = distinct !{[[LOOP12]], [[META4]]}
// CHECK1: [[LOOP13]] = distinct !{[[LOOP13]], [[META4]]}
// CHECK1: [[LOOP14]] = distinct !{[[LOOP14]], [[META4]]}
// CHECK1: [[LOOP15]] = distinct !{[[LOOP15]], [[META4]]}
// CHECK1: [[LOOP16]] = distinct !{[[LOOP16]], [[META4]]}
// CHECK1: [[LOOP17]] = distinct !{[[LOOP17]], [[META4]]}
// CHECK1: [[LOOP18]] = distinct !{[[LOOP18]], [[META4]]}
// CHECK1: [[LOOP21]] = distinct !{[[LOOP21]], [[META4]]}
// CHECK1: [[LOOP22]] = distinct !{[[LOOP22]], [[META4]]}
// CHECK1: [[LOOP23]] = distinct !{[[LOOP23]], [[META4]]}
// CHECK1: [[LOOP24]] = distinct !{[[LOOP24]], [[META4]]}
// CHECK1: [[LOOP25]] = distinct !{[[LOOP25]], [[META4]]}
// CHECK1: [[LOOP26]] = distinct !{[[LOOP26]], [[META4]]}
// CHECK1: [[LOOP27]] = distinct !{[[LOOP27]], [[META4]]}
// CHECK1: [[LOOP28]] = distinct !{[[LOOP28]], [[META4]]}
//.
// CHECK2: [[META2]] = !{}
// CHECK2: [[META3]] = !{i64 4}
// CHECK2: [[LOOP4]] = distinct !{[[LOOP4]], [[META5:![0-9]+]]}
// CHECK2: [[META5]] = !{!"llvm.loop.mustprogress"}
// CHECK2: [[LOOP6]] = distinct !{[[LOOP6]], [[META5]]}
// CHECK2: [[LOOP7]] = distinct !{[[LOOP7]], [[META5]]}
// CHECK2: [[LOOP8]] = distinct !{[[LOOP8]], [[META5]]}
// CHECK2: [[META9]] = !{i64 8}
// CHECK2: [[LOOP10]] = distinct !{[[LOOP10]], [[META5]]}
// CHECK2: [[LOOP11]] = distinct !{[[LOOP11]], [[META5]]}
// CHECK2: [[LOOP12]] = distinct !{[[LOOP12]], [[META5]]}
// CHECK2: [[LOOP13]] = distinct !{[[LOOP13]], [[META5]]}
// CHECK2: [[LOOP14]] = distinct !{[[LOOP14]], [[META5]]}
// CHECK2: [[LOOP15]] = distinct !{[[LOOP15]], [[META5]]}
// CHECK2: [[LOOP16]] = distinct !{[[LOOP16]], [[META5]]}
// CHECK2: [[LOOP17]] = distinct !{[[LOOP17]], [[META5]]}
// CHECK2: [[LOOP18]] = distinct !{[[LOOP18]], [[META5]]}
// CHECK2: [[LOOP19]] = distinct !{[[LOOP19]], [[META5]]}
// CHECK2: [[LOOP20]] = distinct !{[[LOOP20]], [[META5]]}
// CHECK2: [[LOOP21]] = distinct !{[[LOOP21]], [[META5]]}
// CHECK2: [[LOOP22]] = distinct !{[[LOOP22]], [[META5]]}
// CHECK2: [[LOOP25]] = distinct !{[[LOOP25]], [[META5]]}
// CHECK2: [[LOOP26]] = distinct !{[[LOOP26]], [[META5]]}
// CHECK2: [[LOOP27]] = distinct !{[[LOOP27]], [[META5]]}
// CHECK2: [[LOOP28]] = distinct !{[[LOOP28]], [[META5]]}
// CHECK2: [[LOOP29]] = distinct !{[[LOOP29]], [[META5]]}
// CHECK2: [[LOOP30]] = distinct !{[[LOOP30]], [[META5]]}
// CHECK2: [[LOOP3]] = distinct !{[[LOOP3]], [[META4:![0-9]+]]}
// CHECK2: [[META4]] = !{!"llvm.loop.mustprogress"}
// CHECK2: [[LOOP5]] = distinct !{[[LOOP5]], [[META4]]}
// CHECK2: [[LOOP6]] = distinct !{[[LOOP6]], [[META4]]}
// CHECK2: [[LOOP7]] = distinct !{[[LOOP7]], [[META4]]}
// CHECK2: [[LOOP8]] = distinct !{[[LOOP8]], [[META4]]}
// CHECK2: [[LOOP9]] = distinct !{[[LOOP9]], [[META4]]}
// CHECK2: [[LOOP10]] = distinct !{[[LOOP10]], [[META4]]}
// CHECK2: [[LOOP11]] = distinct !{[[LOOP11]], [[META4]]}
// CHECK2: [[LOOP12]] = distinct !{[[LOOP12]], [[META4]]}
// CHECK2: [[LOOP13]] = distinct !{[[LOOP13]], [[META4]]}
// CHECK2: [[LOOP14]] = distinct !{[[LOOP14]], [[META4]]}
// CHECK2: [[LOOP15]] = distinct !{[[LOOP15]], [[META4]]}
// CHECK2: [[LOOP16]] = distinct !{[[LOOP16]], [[META4]]}
// CHECK2: [[LOOP17]] = distinct !{[[LOOP17]], [[META4]]}
// CHECK2: [[LOOP18]] = distinct !{[[LOOP18]], [[META4]]}
// CHECK2: [[LOOP19]] = distinct !{[[LOOP19]], [[META4]]}
// CHECK2: [[LOOP20]] = distinct !{[[LOOP20]], [[META4]]}
// CHECK2: [[LOOP23]] = distinct !{[[LOOP23]], [[META4]]}
// CHECK2: [[LOOP24]] = distinct !{[[LOOP24]], [[META4]]}
// CHECK2: [[LOOP25]] = distinct !{[[LOOP25]], [[META4]]}
// CHECK2: [[LOOP26]] = distinct !{[[LOOP26]], [[META4]]}
// CHECK2: [[LOOP27]] = distinct !{[[LOOP27]], [[META4]]}
// CHECK2: [[LOOP28]] = distinct !{[[LOOP28]], [[META4]]}
//.

282
clang/test/OpenMP/tile_codegen.cpp
View file

@ -174,9 +174,9 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTTILE_0_IV_I]], align 4
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP5]], 3
// CHECK1-NEXT: [[ADD9:%.*]] = add nsw i32 7, [[MUL]]
// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[I2]], align 8, !nonnull [[META2:![0-9]+]], !align [[META3:![0-9]+]]
// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[I2]], align 8
// CHECK1-NEXT: store i32 [[ADD9]], ptr [[TMP6]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[I2]], align 8, !nonnull [[META2]], !align [[META3]]
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[I2]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: call void (...) @body(i32 noundef [[TMP8]])
// CHECK1-NEXT: br label [[FOR_INC:%.*]]
@ -184,17 +184,15 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTTILE_0_IV_I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP9]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTTILE_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND4]], !llvm.loop [[LOOP4:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND4]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC10:%.*]]
// CHECK1: for.inc10:
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: [[ADD11:%.*]] = add nsw i32 [[TMP10]], 5
// CHECK1-NEXT: store i32 [[ADD11]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]]
// CHECK1: for.end12:
// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[I2]], align 8, !nonnull [[META2]], !align [[META3]]
// CHECK1-NEXT: store i32 19, ptr [[TMP11]], align 4
// CHECK1-NEXT: ret void
//
//
@ -278,28 +276,15 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTTILE_0_IV_I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add i32 [[TMP20]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTTILE_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC15:%.*]]
// CHECK1: for.inc15:
// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: [[ADD16:%.*]] = add i32 [[TMP21]], 5
// CHECK1-NEXT: store i32 [[ADD16]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK1: for.end17:
// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[SUB18:%.*]] = sub i32 [[TMP23]], [[TMP24]]
// CHECK1-NEXT: [[SUB19:%.*]] = sub i32 [[SUB18]], 1
// CHECK1-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK1-NEXT: [[ADD20:%.*]] = add i32 [[SUB19]], [[TMP25]]
// CHECK1-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK1-NEXT: [[DIV21:%.*]] = udiv i32 [[ADD20]], [[TMP26]]
// CHECK1-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK1-NEXT: [[MUL22:%.*]] = mul i32 [[DIV21]], [[TMP27]]
// CHECK1-NEXT: [[ADD23:%.*]] = add i32 [[TMP22]], [[MUL22]]
// CHECK1-NEXT: store i32 [[ADD23]], ptr [[I]], align 4
// CHECK1-NEXT: ret void
//
//
@ -390,31 +375,29 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTTILE_1_IV_J]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTTILE_1_IV_J]], align 4
// CHECK1-NEXT: br label [[FOR_COND10]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND10]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC22:%.*]]
// CHECK1: for.inc22:
// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTTILE_0_IV_I]], align 4
// CHECK1-NEXT: [[INC23:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK1-NEXT: store i32 [[INC23]], ptr [[DOTTILE_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND4]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND4]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK1: for.end24:
// CHECK1-NEXT: br label [[FOR_INC25:%.*]]
// CHECK1: for.inc25:
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK1-NEXT: [[ADD26:%.*]] = add nsw i32 [[TMP16]], 5
// CHECK1-NEXT: store i32 [[ADD26]], ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK1-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK1: for.end27:
// CHECK1-NEXT: br label [[FOR_INC28:%.*]]
// CHECK1: for.inc28:
// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP17]], 5
// CHECK1-NEXT: store i32 [[ADD29]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK1: for.end30:
// CHECK1-NEXT: store i32 19, ptr [[I]], align 4
// CHECK1-NEXT: store i32 19, ptr [[J]], align 4
// CHECK1-NEXT: ret void
//
//
@ -528,21 +511,21 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTTILE_1_IV_J]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP20]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTTILE_1_IV_J]], align 4
// CHECK1-NEXT: br label [[FOR_COND15]], !llvm.loop [[LOOP13:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND15]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC27:%.*]]
// CHECK1: for.inc27:
// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTTILE_0_IV_I]], align 4
// CHECK1-NEXT: [[INC28:%.*]] = add nsw i32 [[TMP21]], 1
// CHECK1-NEXT: store i32 [[INC28]], ptr [[DOTTILE_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND3]], !llvm.loop [[LOOP14:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND3]], !llvm.loop [[LOOP13:![0-9]+]]
// CHECK1: for.end29:
// CHECK1-NEXT: br label [[FOR_INC30:%.*]]
// CHECK1: for.inc30:
// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK1-NEXT: [[ADD31:%.*]] = add nsw i32 [[TMP22]], 5
// CHECK1-NEXT: store i32 [[ADD31]], ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]]
// CHECK1: for.end32:
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
@ -681,21 +664,21 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTTILE_1_IV_J]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP22]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTTILE_1_IV_J]], align 4
// CHECK1-NEXT: br label [[FOR_COND20]], !llvm.loop [[LOOP16:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND20]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC32:%.*]]
// CHECK1: for.inc32:
// CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTTILE_0_IV_I]], align 4
// CHECK1-NEXT: [[INC33:%.*]] = add nsw i32 [[TMP23]], 1
// CHECK1-NEXT: store i32 [[INC33]], ptr [[DOTTILE_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND8]], !llvm.loop [[LOOP17:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND8]], !llvm.loop [[LOOP16:![0-9]+]]
// CHECK1: for.end34:
// CHECK1-NEXT: br label [[FOR_INC35:%.*]]
// CHECK1: for.inc35:
// CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK1-NEXT: [[ADD36:%.*]] = add nsw i32 [[TMP24]], 5
// CHECK1-NEXT: store i32 [[ADD36]], ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP18:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP17:![0-9]+]]
// CHECK1: for.end37:
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
@ -1008,7 +991,7 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTTILE_0_IV_I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTTILE_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP18:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
@ -1104,25 +1087,15 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTTILE_0_IV_I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add i32 [[TMP16]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTTILE_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP22:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP21:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC15:%.*]]
// CHECK1: for.inc15:
// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: [[ADD16:%.*]] = add i32 [[TMP17]], 5
// CHECK1-NEXT: store i32 [[ADD16]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP23:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP22:![0-9]+]]
// CHECK1: for.end17:
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK1-NEXT: [[SUB18:%.*]] = sub i32 [[TMP19]], [[TMP20]]
// CHECK1-NEXT: [[SUB19:%.*]] = sub i32 [[SUB18]], 1
// CHECK1-NEXT: [[ADD20:%.*]] = add i32 [[SUB19]], 3
// CHECK1-NEXT: [[DIV21:%.*]] = udiv i32 [[ADD20]], 3
// CHECK1-NEXT: [[MUL22:%.*]] = mul i32 [[DIV21]], 3
// CHECK1-NEXT: [[ADD23:%.*]] = add i32 [[TMP18]], [[MUL22]]
// CHECK1-NEXT: store i32 [[ADD23]], ptr [[I]], align 4
// CHECK1-NEXT: ret void
//
//
@ -1193,7 +1166,7 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTTILE_0_IV_I]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP11]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[DOTTILE_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP24:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP23:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC17:%.*]]
// CHECK1: for.inc17:
@ -1210,9 +1183,8 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: [[ADD23:%.*]] = add nsw i32 [[TMP14]], [[COND22]]
// CHECK1-NEXT: store i32 [[ADD23]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP25:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP24:![0-9]+]]
// CHECK1: for.end24:
// CHECK1-NEXT: store i32 19, ptr [[I]], align 4
// CHECK1-NEXT: ret void
//
//
@ -1230,14 +1202,14 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[V:%.*]] = alloca double, align 8
// CHECK1-NEXT: [[ARRAY:%.*]] = getelementptr inbounds nuw [[STRUCT_DATA_T]], ptr [[DATA]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[ARRAY]], ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META26:![0-9]+]]
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP0]], i64 0, i64 0
// CHECK1-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 12
// CHECK1-NEXT: store ptr [[ADD_PTR]], ptr [[__END2]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META26]]
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY1:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP1]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY1]], ptr [[__BEGIN2]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META26]]
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY2:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY2]], ptr [[DOTCAPTURE_EXPR_]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__END2]], align 8
@ -1301,14 +1273,14 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTTILE_0_IV___BEGIN2]], align 8
// CHECK1-NEXT: [[INC:%.*]] = add nsw i64 [[TMP19]], 1
// CHECK1-NEXT: store i64 [[INC]], ptr [[DOTTILE_0_IV___BEGIN2]], align 8
// CHECK1-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP27:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP25:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC16:%.*]]
// CHECK1: for.inc16:
// CHECK1-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTFLOOR_0_IV___BEGIN2]], align 8
// CHECK1-NEXT: [[ADD17:%.*]] = add nsw i64 [[TMP20]], 5
// CHECK1-NEXT: store i64 [[ADD17]], ptr [[DOTFLOOR_0_IV___BEGIN2]], align 8
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP28:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP26:![0-9]+]]
// CHECK1: for.end18:
// CHECK1-NEXT: ret void
//
@ -1329,14 +1301,14 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: store double 4.200000e+01, ptr [[C]], align 8
// CHECK1-NEXT: [[ARRAY:%.*]] = getelementptr inbounds nuw [[STRUCT_DATA_T]], ptr [[DATA]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[ARRAY]], ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META26]]
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP0]], i64 0, i64 0
// CHECK1-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 12
// CHECK1-NEXT: store ptr [[ADD_PTR]], ptr [[__END2]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META26]]
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY1:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP1]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY1]], ptr [[__BEGIN2]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META26]]
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK1-NEXT: [[ARRAYDECAY2:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARRAYDECAY2]], ptr [[DOTCAPTURE_EXPR_]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__END2]], align 8
@ -1401,14 +1373,14 @@ extern "C" void foo10(data_t data) {
// CHECK1-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTTILE_0_IV___BEGIN2]], align 8
// CHECK1-NEXT: [[INC:%.*]] = add nsw i64 [[TMP20]], 1
// CHECK1-NEXT: store i64 [[INC]], ptr [[DOTTILE_0_IV___BEGIN2]], align 8
// CHECK1-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP29:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP27:![0-9]+]]
// CHECK1: for.end:
// CHECK1-NEXT: br label [[FOR_INC16:%.*]]
// CHECK1: for.inc16:
// CHECK1-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTFLOOR_0_IV___BEGIN2]], align 8
// CHECK1-NEXT: [[ADD17:%.*]] = add nsw i64 [[TMP21]], 5
// CHECK1-NEXT: store i64 [[ADD17]], ptr [[DOTFLOOR_0_IV___BEGIN2]], align 8
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP30:![0-9]+]]
// CHECK1-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP28:![0-9]+]]
// CHECK1: for.end18:
// CHECK1-NEXT: ret void
//
@ -1480,9 +1452,9 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTTILE_0_IV_I]], align 4
// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP5]], 3
// CHECK2-NEXT: [[ADD9:%.*]] = add nsw i32 7, [[MUL]]
// CHECK2-NEXT: [[TMP6:%.*]] = load ptr, ptr [[I2]], align 8, !nonnull [[META2:![0-9]+]], !align [[META3:![0-9]+]]
// CHECK2-NEXT: [[TMP6:%.*]] = load ptr, ptr [[I2]], align 8
// CHECK2-NEXT: store i32 [[ADD9]], ptr [[TMP6]], align 4
// CHECK2-NEXT: [[TMP7:%.*]] = load ptr, ptr [[I2]], align 8, !nonnull [[META2]], !align [[META3]]
// CHECK2-NEXT: [[TMP7:%.*]] = load ptr, ptr [[I2]], align 8
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK2-NEXT: call void (...) @body(i32 noundef [[TMP8]])
// CHECK2-NEXT: br label [[FOR_INC:%.*]]
@ -1490,17 +1462,15 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTTILE_0_IV_I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP9]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTTILE_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND4]], !llvm.loop [[LOOP4:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND4]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC10:%.*]]
// CHECK2: for.inc10:
// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: [[ADD11:%.*]] = add nsw i32 [[TMP10]], 5
// CHECK2-NEXT: store i32 [[ADD11]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]]
// CHECK2: for.end12:
// CHECK2-NEXT: [[TMP11:%.*]] = load ptr, ptr [[I2]], align 8, !nonnull [[META2]], !align [[META3]]
// CHECK2-NEXT: store i32 19, ptr [[TMP11]], align 4
// CHECK2-NEXT: ret void
//
//
@ -1590,28 +1560,15 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTTILE_0_IV_I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add i32 [[TMP20]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTTILE_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC15:%.*]]
// CHECK2: for.inc15:
// CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: [[ADD16:%.*]] = add i32 [[TMP21]], 5
// CHECK2-NEXT: store i32 [[ADD16]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK2: for.end17:
// CHECK2-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK2-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[SUB18:%.*]] = sub i32 [[TMP23]], [[TMP24]]
// CHECK2-NEXT: [[SUB19:%.*]] = sub i32 [[SUB18]], 1
// CHECK2-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK2-NEXT: [[ADD20:%.*]] = add i32 [[SUB19]], [[TMP25]]
// CHECK2-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK2-NEXT: [[DIV21:%.*]] = udiv i32 [[ADD20]], [[TMP26]]
// CHECK2-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTNEW_STEP]], align 4
// CHECK2-NEXT: [[MUL22:%.*]] = mul i32 [[DIV21]], [[TMP27]]
// CHECK2-NEXT: [[ADD23:%.*]] = add i32 [[TMP22]], [[MUL22]]
// CHECK2-NEXT: store i32 [[ADD23]], ptr [[I]], align 4
// CHECK2-NEXT: ret void
//
//
@ -1631,14 +1588,14 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: store double 4.200000e+01, ptr [[C]], align 8
// CHECK2-NEXT: [[ARRAY:%.*]] = getelementptr inbounds nuw [[STRUCT_DATA_T]], ptr [[DATA]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[ARRAY]], ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META9:![0-9]+]]
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP0]], i64 0, i64 0
// CHECK2-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 12
// CHECK2-NEXT: store ptr [[ADD_PTR]], ptr [[__END2]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META9]]
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY1:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP1]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY1]], ptr [[__BEGIN2]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META9]]
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY2:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP2]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY2]], ptr [[DOTCAPTURE_EXPR_]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__END2]], align 8
@ -1703,14 +1660,14 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTTILE_0_IV___BEGIN2]], align 8
// CHECK2-NEXT: [[INC:%.*]] = add nsw i64 [[TMP20]], 1
// CHECK2-NEXT: store i64 [[INC]], ptr [[DOTTILE_0_IV___BEGIN2]], align 8
// CHECK2-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC16:%.*]]
// CHECK2: for.inc16:
// CHECK2-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTFLOOR_0_IV___BEGIN2]], align 8
// CHECK2-NEXT: [[ADD17:%.*]] = add nsw i64 [[TMP21]], 5
// CHECK2-NEXT: store i64 [[ADD17]], ptr [[DOTFLOOR_0_IV___BEGIN2]], align 8
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK2: for.end18:
// CHECK2-NEXT: ret void
//
@ -1802,31 +1759,29 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTTILE_1_IV_J]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTTILE_1_IV_J]], align 4
// CHECK2-NEXT: br label [[FOR_COND10]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND10]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC22:%.*]]
// CHECK2: for.inc22:
// CHECK2-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTTILE_0_IV_I]], align 4
// CHECK2-NEXT: [[INC23:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK2-NEXT: store i32 [[INC23]], ptr [[DOTTILE_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND4]], !llvm.loop [[LOOP13:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND4]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK2: for.end24:
// CHECK2-NEXT: br label [[FOR_INC25:%.*]]
// CHECK2: for.inc25:
// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK2-NEXT: [[ADD26:%.*]] = add nsw i32 [[TMP16]], 5
// CHECK2-NEXT: store i32 [[ADD26]], ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK2-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP14:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK2: for.end27:
// CHECK2-NEXT: br label [[FOR_INC28:%.*]]
// CHECK2: for.inc28:
// CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP17]], 5
// CHECK2-NEXT: store i32 [[ADD29]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP13:![0-9]+]]
// CHECK2: for.end30:
// CHECK2-NEXT: store i32 19, ptr [[I]], align 4
// CHECK2-NEXT: store i32 19, ptr [[J]], align 4
// CHECK2-NEXT: ret void
//
//
@ -1940,21 +1895,21 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTTILE_1_IV_J]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP20]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTTILE_1_IV_J]], align 4
// CHECK2-NEXT: br label [[FOR_COND15]], !llvm.loop [[LOOP16:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND15]], !llvm.loop [[LOOP14:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC27:%.*]]
// CHECK2: for.inc27:
// CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTTILE_0_IV_I]], align 4
// CHECK2-NEXT: [[INC28:%.*]] = add nsw i32 [[TMP21]], 1
// CHECK2-NEXT: store i32 [[INC28]], ptr [[DOTTILE_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND3]], !llvm.loop [[LOOP17:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND3]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK2: for.end29:
// CHECK2-NEXT: br label [[FOR_INC30:%.*]]
// CHECK2: for.inc30:
// CHECK2-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK2-NEXT: [[ADD31:%.*]] = add nsw i32 [[TMP22]], 5
// CHECK2-NEXT: store i32 [[ADD31]], ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP18:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]]
// CHECK2: for.end32:
// CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2: omp.body.continue:
@ -2093,21 +2048,21 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTTILE_1_IV_J]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP22]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTTILE_1_IV_J]], align 4
// CHECK2-NEXT: br label [[FOR_COND20]], !llvm.loop [[LOOP19:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND20]], !llvm.loop [[LOOP17:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC32:%.*]]
// CHECK2: for.inc32:
// CHECK2-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTTILE_0_IV_I]], align 4
// CHECK2-NEXT: [[INC33:%.*]] = add nsw i32 [[TMP23]], 1
// CHECK2-NEXT: store i32 [[INC33]], ptr [[DOTTILE_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND8]], !llvm.loop [[LOOP20:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND8]], !llvm.loop [[LOOP18:![0-9]+]]
// CHECK2: for.end34:
// CHECK2-NEXT: br label [[FOR_INC35:%.*]]
// CHECK2: for.inc35:
// CHECK2-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK2-NEXT: [[ADD36:%.*]] = add nsw i32 [[TMP24]], 5
// CHECK2-NEXT: store i32 [[ADD36]], ptr [[DOTFLOOR_1_IV_J]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP21:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]]
// CHECK2: for.end37:
// CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2: omp.body.continue:
@ -2420,7 +2375,7 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTTILE_0_IV_I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTTILE_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP22:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP20:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2: omp.body.continue:
@ -2504,7 +2459,7 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTTILE_0_IV_I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP11]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTTILE_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP25:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND1]], !llvm.loop [[LOOP23:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC17:%.*]]
// CHECK2: for.inc17:
@ -2521,9 +2476,8 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: [[ADD23:%.*]] = add nsw i32 [[TMP14]], [[COND22]]
// CHECK2-NEXT: store i32 [[ADD23]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP26:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP24:![0-9]+]]
// CHECK2: for.end24:
// CHECK2-NEXT: store i32 19, ptr [[I]], align 4
// CHECK2-NEXT: ret void
//
//
@ -2541,14 +2495,14 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[V:%.*]] = alloca double, align 8
// CHECK2-NEXT: [[ARRAY:%.*]] = getelementptr inbounds nuw [[STRUCT_DATA_T]], ptr [[DATA]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[ARRAY]], ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META9]]
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP0]], i64 0, i64 0
// CHECK2-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds double, ptr [[ARRAYDECAY]], i64 12
// CHECK2-NEXT: store ptr [[ADD_PTR]], ptr [[__END2]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META9]]
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY1:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP1]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY1]], ptr [[__BEGIN2]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8, !nonnull [[META2]], !align [[META9]]
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[__RANGE2]], align 8
// CHECK2-NEXT: [[ARRAYDECAY2:%.*]] = getelementptr inbounds [12 x double], ptr [[TMP2]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[ARRAYDECAY2]], ptr [[DOTCAPTURE_EXPR_]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[__END2]], align 8
@ -2612,14 +2566,14 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTTILE_0_IV___BEGIN2]], align 8
// CHECK2-NEXT: [[INC:%.*]] = add nsw i64 [[TMP19]], 1
// CHECK2-NEXT: store i64 [[INC]], ptr [[DOTTILE_0_IV___BEGIN2]], align 8
// CHECK2-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP27:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND7]], !llvm.loop [[LOOP25:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC16:%.*]]
// CHECK2: for.inc16:
// CHECK2-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTFLOOR_0_IV___BEGIN2]], align 8
// CHECK2-NEXT: [[ADD17:%.*]] = add nsw i64 [[TMP20]], 5
// CHECK2-NEXT: store i64 [[ADD17]], ptr [[DOTFLOOR_0_IV___BEGIN2]], align 8
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP28:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP26:![0-9]+]]
// CHECK2: for.end18:
// CHECK2-NEXT: ret void
//
@ -2703,25 +2657,15 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTTILE_0_IV_I]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add i32 [[TMP16]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[DOTTILE_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP29:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND6]], !llvm.loop [[LOOP27:![0-9]+]]
// CHECK2: for.end:
// CHECK2-NEXT: br label [[FOR_INC15:%.*]]
// CHECK2: for.inc15:
// CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: [[ADD16:%.*]] = add i32 [[TMP17]], 5
// CHECK2-NEXT: store i32 [[ADD16]], ptr [[DOTFLOOR_0_IV_I]], align 4
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP30:![0-9]+]]
// CHECK2-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP28:![0-9]+]]
// CHECK2: for.end17:
// CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK2-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK2-NEXT: [[SUB18:%.*]] = sub i32 [[TMP19]], [[TMP20]]
// CHECK2-NEXT: [[SUB19:%.*]] = sub i32 [[SUB18]], 1
// CHECK2-NEXT: [[ADD20:%.*]] = add i32 [[SUB19]], 3
// CHECK2-NEXT: [[DIV21:%.*]] = udiv i32 [[ADD20]], 3
// CHECK2-NEXT: [[MUL22:%.*]] = mul i32 [[DIV21]], 3
// CHECK2-NEXT: [[ADD23:%.*]] = add i32 [[TMP18]], [[MUL22]]
// CHECK2-NEXT: store i32 [[ADD23]], ptr [[I]], align 4
// CHECK2-NEXT: ret void
//
//
@ -2732,59 +2676,53 @@ extern "C" void foo10(data_t data) {
// CHECK2-NEXT: ret void
//
//.
// CHECK1: [[META2]] = !{}
// CHECK1: [[META3]] = !{i64 4}
// CHECK1: [[LOOP4]] = distinct !{[[LOOP4]], [[META5:![0-9]+]]}
// CHECK1: [[META5]] = !{!"llvm.loop.mustprogress"}
// CHECK1: [[LOOP6]] = distinct !{[[LOOP6]], [[META5]]}
// CHECK1: [[LOOP7]] = distinct !{[[LOOP7]], [[META5]]}
// CHECK1: [[LOOP8]] = distinct !{[[LOOP8]], [[META5]]}
// CHECK1: [[LOOP9]] = distinct !{[[LOOP9]], [[META5]]}
// CHECK1: [[LOOP10]] = distinct !{[[LOOP10]], [[META5]]}
// CHECK1: [[LOOP11]] = distinct !{[[LOOP11]], [[META5]]}
// CHECK1: [[LOOP12]] = distinct !{[[LOOP12]], [[META5]]}
// CHECK1: [[LOOP13]] = distinct !{[[LOOP13]], [[META5]]}
// CHECK1: [[LOOP14]] = distinct !{[[LOOP14]], [[META5]]}
// CHECK1: [[LOOP15]] = distinct !{[[LOOP15]], [[META5]]}
// CHECK1: [[LOOP16]] = distinct !{[[LOOP16]], [[META5]]}
// CHECK1: [[LOOP17]] = distinct !{[[LOOP17]], [[META5]]}
// CHECK1: [[LOOP18]] = distinct !{[[LOOP18]], [[META5]]}
// CHECK1: [[LOOP19]] = distinct !{[[LOOP19]], [[META5]]}
// CHECK1: [[LOOP22]] = distinct !{[[LOOP22]], [[META5]]}
// CHECK1: [[LOOP23]] = distinct !{[[LOOP23]], [[META5]]}
// CHECK1: [[LOOP24]] = distinct !{[[LOOP24]], [[META5]]}
// CHECK1: [[LOOP25]] = distinct !{[[LOOP25]], [[META5]]}
// CHECK1: [[META26]] = !{i64 8}
// CHECK1: [[LOOP27]] = distinct !{[[LOOP27]], [[META5]]}
// CHECK1: [[LOOP28]] = distinct !{[[LOOP28]], [[META5]]}
// CHECK1: [[LOOP29]] = distinct !{[[LOOP29]], [[META5]]}
// CHECK1: [[LOOP30]] = distinct !{[[LOOP30]], [[META5]]}
// CHECK1: [[LOOP3]] = distinct !{[[LOOP3]], [[META4:![0-9]+]]}
// CHECK1: [[META4]] = !{!"llvm.loop.mustprogress"}
// CHECK1: [[LOOP5]] = distinct !{[[LOOP5]], [[META4]]}
// CHECK1: [[LOOP6]] = distinct !{[[LOOP6]], [[META4]]}
// CHECK1: [[LOOP7]] = distinct !{[[LOOP7]], [[META4]]}
// CHECK1: [[LOOP8]] = distinct !{[[LOOP8]], [[META4]]}
// CHECK1: [[LOOP9]] = distinct !{[[LOOP9]], [[META4]]}
// CHECK1: [[LOOP10]] = distinct !{[[LOOP10]], [[META4]]}
// CHECK1: [[LOOP11]] = distinct !{[[LOOP11]], [[META4]]}
// CHECK1: [[LOOP12]] = distinct !{[[LOOP12]], [[META4]]}
// CHECK1: [[LOOP13]] = distinct !{[[LOOP13]], [[META4]]}
// CHECK1: [[LOOP14]] = distinct !{[[LOOP14]], [[META4]]}
// CHECK1: [[LOOP15]] = distinct !{[[LOOP15]], [[META4]]}
// CHECK1: [[LOOP16]] = distinct !{[[LOOP16]], [[META4]]}
// CHECK1: [[LOOP17]] = distinct !{[[LOOP17]], [[META4]]}
// CHECK1: [[LOOP18]] = distinct !{[[LOOP18]], [[META4]]}
// CHECK1: [[LOOP21]] = distinct !{[[LOOP21]], [[META4]]}
// CHECK1: [[LOOP22]] = distinct !{[[LOOP22]], [[META4]]}
// CHECK1: [[LOOP23]] = distinct !{[[LOOP23]], [[META4]]}
// CHECK1: [[LOOP24]] = distinct !{[[LOOP24]], [[META4]]}
// CHECK1: [[LOOP25]] = distinct !{[[LOOP25]], [[META4]]}
// CHECK1: [[LOOP26]] = distinct !{[[LOOP26]], [[META4]]}
// CHECK1: [[LOOP27]] = distinct !{[[LOOP27]], [[META4]]}
// CHECK1: [[LOOP28]] = distinct !{[[LOOP28]], [[META4]]}
//.
// CHECK2: [[META2]] = !{}
// CHECK2: [[META3]] = !{i64 4}
// CHECK2: [[LOOP4]] = distinct !{[[LOOP4]], [[META5:![0-9]+]]}
// CHECK2: [[META5]] = !{!"llvm.loop.mustprogress"}
// CHECK2: [[LOOP6]] = distinct !{[[LOOP6]], [[META5]]}
// CHECK2: [[LOOP7]] = distinct !{[[LOOP7]], [[META5]]}
// CHECK2: [[LOOP8]] = distinct !{[[LOOP8]], [[META5]]}
// CHECK2: [[META9]] = !{i64 8}
// CHECK2: [[LOOP10]] = distinct !{[[LOOP10]], [[META5]]}
// CHECK2: [[LOOP11]] = distinct !{[[LOOP11]], [[META5]]}
// CHECK2: [[LOOP12]] = distinct !{[[LOOP12]], [[META5]]}
// CHECK2: [[LOOP13]] = distinct !{[[LOOP13]], [[META5]]}
// CHECK2: [[LOOP14]] = distinct !{[[LOOP14]], [[META5]]}
// CHECK2: [[LOOP15]] = distinct !{[[LOOP15]], [[META5]]}
// CHECK2: [[LOOP16]] = distinct !{[[LOOP16]], [[META5]]}
// CHECK2: [[LOOP17]] = distinct !{[[LOOP17]], [[META5]]}
// CHECK2: [[LOOP18]] = distinct !{[[LOOP18]], [[META5]]}
// CHECK2: [[LOOP19]] = distinct !{[[LOOP19]], [[META5]]}
// CHECK2: [[LOOP20]] = distinct !{[[LOOP20]], [[META5]]}
// CHECK2: [[LOOP21]] = distinct !{[[LOOP21]], [[META5]]}
// CHECK2: [[LOOP22]] = distinct !{[[LOOP22]], [[META5]]}
// CHECK2: [[LOOP25]] = distinct !{[[LOOP25]], [[META5]]}
// CHECK2: [[LOOP26]] = distinct !{[[LOOP26]], [[META5]]}
// CHECK2: [[LOOP27]] = distinct !{[[LOOP27]], [[META5]]}
// CHECK2: [[LOOP28]] = distinct !{[[LOOP28]], [[META5]]}
// CHECK2: [[LOOP29]] = distinct !{[[LOOP29]], [[META5]]}
// CHECK2: [[LOOP30]] = distinct !{[[LOOP30]], [[META5]]}
// CHECK2: [[LOOP3]] = distinct !{[[LOOP3]], [[META4:![0-9]+]]}
// CHECK2: [[META4]] = !{!"llvm.loop.mustprogress"}
// CHECK2: [[LOOP5]] = distinct !{[[LOOP5]], [[META4]]}
// CHECK2: [[LOOP6]] = distinct !{[[LOOP6]], [[META4]]}
// CHECK2: [[LOOP7]] = distinct !{[[LOOP7]], [[META4]]}
// CHECK2: [[LOOP8]] = distinct !{[[LOOP8]], [[META4]]}
// CHECK2: [[LOOP9]] = distinct !{[[LOOP9]], [[META4]]}
// CHECK2: [[LOOP10]] = distinct !{[[LOOP10]], [[META4]]}
// CHECK2: [[LOOP11]] = distinct !{[[LOOP11]], [[META4]]}
// CHECK2: [[LOOP12]] = distinct !{[[LOOP12]], [[META4]]}
// CHECK2: [[LOOP13]] = distinct !{[[LOOP13]], [[META4]]}
// CHECK2: [[LOOP14]] = distinct !{[[LOOP14]], [[META4]]}
// CHECK2: [[LOOP15]] = distinct !{[[LOOP15]], [[META4]]}
// CHECK2: [[LOOP16]] = distinct !{[[LOOP16]], [[META4]]}
// CHECK2: [[LOOP17]] = distinct !{[[LOOP17]], [[META4]]}
// CHECK2: [[LOOP18]] = distinct !{[[LOOP18]], [[META4]]}
// CHECK2: [[LOOP19]] = distinct !{[[LOOP19]], [[META4]]}
// CHECK2: [[LOOP20]] = distinct !{[[LOOP20]], [[META4]]}
// CHECK2: [[LOOP23]] = distinct !{[[LOOP23]], [[META4]]}
// CHECK2: [[LOOP24]] = distinct !{[[LOOP24]], [[META4]]}
// CHECK2: [[LOOP25]] = distinct !{[[LOOP25]], [[META4]]}
// CHECK2: [[LOOP26]] = distinct !{[[LOOP26]], [[META4]]}
// CHECK2: [[LOOP27]] = distinct !{[[LOOP27]], [[META4]]}
// CHECK2: [[LOOP28]] = distinct !{[[LOOP28]], [[META4]]}
//.