[clang] Reapply "[clang][CodeGen] Zero init unspecified fields in initializers in C" (#109898) (PR #110051)
via cfe-commits
cfe-commits at lists.llvm.org
Fri Oct 4 15:14:03 PDT 2024
https://github.com/yabinc updated https://github.com/llvm/llvm-project/pull/110051
>From 112b018d3cac542757fcc399efa9fa5a9caac9b8 Mon Sep 17 00:00:00 2001
From: Yabin Cui <yabinc at google.com>
Date: Wed, 25 Sep 2024 10:05:27 -0700
Subject: [PATCH 1/3] Reapply "[clang][CodeGen] Zero init unspecified fields in
initializers in C" (#109898)
This reverts commit d50eaac12f0cdfe27e942290942b06889ab12a8c.
Also fixes a bug calculating offsets for bit fields in the original patch.
---
clang/docs/LanguageExtensions.rst | 23 ++
clang/lib/CodeGen/CGExprAgg.cpp | 52 +++-
clang/lib/CodeGen/CGExprConstant.cpp | 107 ++++++-
clang/lib/CodeGen/CodeGenModule.h | 51 ++++
...07-22-bitfield-init-after-zero-len-array.c | 2 +-
clang/test/CodeGen/2008-08-07-AlignPadding1.c | 4 +-
.../CodeGen/2009-06-14-anonymous-union-init.c | 4 +-
clang/test/CodeGen/64bit-swiftcall.c | 12 +-
clang/test/CodeGen/arm-swiftcall.c | 4 +-
clang/test/CodeGen/const-init.c | 4 +-
clang/test/CodeGen/decl.c | 4 +-
clang/test/CodeGen/designated-initializers.c | 12 +-
clang/test/CodeGen/ext-int.c | 18 +-
clang/test/CodeGen/flexible-array-init.c | 24 +-
clang/test/CodeGen/global-init.c | 2 +-
clang/test/CodeGen/init.c | 19 --
.../linux-kernel-struct-union-initializer.c | 275 ++++++++++++++++++
.../linux-kernel-struct-union-initializer2.c | 178 ++++++++++++
clang/test/CodeGen/mingw-long-double.c | 9 +-
clang/test/CodeGen/mms-bitfields.c | 4 +-
clang/test/CodeGen/union-init2.c | 4 +-
clang/test/CodeGen/windows-swiftcall.c | 12 +-
.../CodeGenObjC/designated-initializers.m | 2 +-
23 files changed, 730 insertions(+), 96 deletions(-)
create mode 100644 clang/test/CodeGen/linux-kernel-struct-union-initializer.c
create mode 100644 clang/test/CodeGen/linux-kernel-struct-union-initializer2.c
diff --git a/clang/docs/LanguageExtensions.rst b/clang/docs/LanguageExtensions.rst
index 0c6b9b1b8f9ce4..f4be97047422fa 100644
--- a/clang/docs/LanguageExtensions.rst
+++ b/clang/docs/LanguageExtensions.rst
@@ -5860,3 +5860,26 @@ specify the starting offset to begin embedding from. The resources is treated
as being empty if the specified offset is larger than the number of bytes in
the resource. The offset will be applied *before* any ``limit`` parameters are
applied.
+
+Union and aggregate initialization in C
+=======================================
+
+In C23 (N2900), when an object is initialized from initializer ``= {}``, all
+elements of arrays, all members of structs, and the first members of unions are
+empty-initialized recursively. In addition, all padding bits are initialized to
+zero.
+
+Clang guarantees the following behaviors:
+
+* ``1:`` Clang supports initializer ``= {}`` mentioned above in all C
+ standards.
+
+* ``2:`` When unions are initialized from initializer ``= {}``, bytes outside
+ of the first members of unions are also initialized to zero.
+
+* ``3:`` When unions, structures and arrays are initialized from initializer
+ ``= { initializer-list }``, all members not explicitly initialized in
+ the initializer list are empty-initialized recursively. In addition, all
+ padding bits are initialized to zero.
+
+Currently, the above extension only applies to C source code, not C++.
diff --git a/clang/lib/CodeGen/CGExprAgg.cpp b/clang/lib/CodeGen/CGExprAgg.cpp
index bbfc6672ecc25a..2bf67cee5fa926 100644
--- a/clang/lib/CodeGen/CGExprAgg.cpp
+++ b/clang/lib/CodeGen/CGExprAgg.cpp
@@ -12,6 +12,7 @@
#include "CGCXXABI.h"
#include "CGObjCRuntime.h"
+#include "CGRecordLayout.h"
#include "CodeGenFunction.h"
#include "CodeGenModule.h"
#include "ConstantEmitter.h"
@@ -1698,6 +1699,17 @@ void AggExprEmitter::VisitCXXParenListOrInitListExpr(
// Prepare a 'this' for CXXDefaultInitExprs.
CodeGenFunction::FieldConstructionScope FCS(CGF, Dest.getAddress());
+ const bool ZeroInitPadding =
+ CGF.CGM.shouldZeroInitPadding() && !Dest.isZeroed();
+ const Address BaseLoc = Dest.getAddress().withElementType(CGF.Int8Ty);
+ auto DoZeroInitPadding = [&](CharUnits Offset, CharUnits Size) {
+ if (Size.isPositive()) {
+ Address Loc = CGF.Builder.CreateConstGEP(BaseLoc, Offset.getQuantity());
+ llvm::Constant *SizeVal = CGF.Builder.getInt64(Size.getQuantity());
+ CGF.Builder.CreateMemSet(Loc, CGF.Builder.getInt8(0), SizeVal, false);
+ }
+ };
+
if (record->isUnion()) {
// Only initialize one field of a union. The field itself is
// specified by the initializer list.
@@ -1722,16 +1734,27 @@ void AggExprEmitter::VisitCXXParenListOrInitListExpr(
if (NumInitElements) {
// Store the initializer into the field
EmitInitializationToLValue(InitExprs[0], FieldLoc);
+ if (ZeroInitPadding) {
+ CharUnits TotalSize =
+ Dest.getPreferredSize(CGF.getContext(), DestLV.getType());
+ CharUnits FieldSize =
+ CGF.getContext().getTypeSizeInChars(FieldLoc.getType());
+ DoZeroInitPadding(FieldSize, TotalSize - FieldSize);
+ }
} else {
// Default-initialize to null.
- EmitNullInitializationToLValue(FieldLoc);
+ if (ZeroInitPadding)
+ EmitNullInitializationToLValue(DestLV);
+ else
+ EmitNullInitializationToLValue(FieldLoc);
}
-
return;
}
// Here we iterate over the fields; this makes it simpler to both
// default-initialize fields and skip over unnamed fields.
+ const ASTRecordLayout &Layout = CGF.getContext().getASTRecordLayout(record);
+ CharUnits SizeSoFar = CharUnits::Zero();
for (const auto *field : record->fields()) {
// We're done once we hit the flexible array member.
if (field->getType()->isIncompleteArrayType())
@@ -1748,6 +1771,26 @@ void AggExprEmitter::VisitCXXParenListOrInitListExpr(
CGF.getTypes().isZeroInitializable(ExprToVisit->getType()))
break;
+ if (ZeroInitPadding) {
+ uint64_t StartBitOffset = Layout.getFieldOffset(field->getFieldIndex());
+ CharUnits StartOffset =
+ CGF.getContext().toCharUnitsFromBits(StartBitOffset);
+ DoZeroInitPadding(SizeSoFar, StartOffset - SizeSoFar);
+ if (!field->isBitField()) {
+ CharUnits FieldSize =
+ CGF.getContext().getTypeSizeInChars(field->getType());
+ SizeSoFar = StartOffset + FieldSize;
+ } else {
+ const CGRecordLayout &RL =
+ CGF.getTypes().getCGRecordLayout(field->getParent());
+ const CGBitFieldInfo &Info = RL.getBitFieldInfo(field);
+ uint64_t EndBitOffset = StartBitOffset + Info.Size;
+ SizeSoFar = CGF.getContext().toCharUnitsFromBits(EndBitOffset);
+ if (EndBitOffset % CGF.getContext().getCharWidth() != 0) {
+ SizeSoFar++;
+ }
+ }
+ }
LValue LV = CGF.EmitLValueForFieldInitialization(DestLV, field);
// We never generate write-barries for initialized fields.
@@ -1774,6 +1817,11 @@ void AggExprEmitter::VisitCXXParenListOrInitListExpr(
}
}
}
+ if (ZeroInitPadding) {
+ CharUnits TotalSize =
+ Dest.getPreferredSize(CGF.getContext(), DestLV.getType());
+ DoZeroInitPadding(SizeSoFar, TotalSize - SizeSoFar);
+ }
}
void AggExprEmitter::VisitArrayInitLoopExpr(const ArrayInitLoopExpr *E,
diff --git a/clang/lib/CodeGen/CGExprConstant.cpp b/clang/lib/CodeGen/CGExprConstant.cpp
index dd65080a840446..655fc3dc954c81 100644
--- a/clang/lib/CodeGen/CGExprConstant.cpp
+++ b/clang/lib/CodeGen/CGExprConstant.cpp
@@ -42,6 +42,16 @@ using namespace CodeGen;
namespace {
class ConstExprEmitter;
+llvm::Constant *getPadding(const CodeGenModule &CGM, CharUnits PadSize) {
+ llvm::Type *Ty = CGM.CharTy;
+ if (PadSize > CharUnits::One())
+ Ty = llvm::ArrayType::get(Ty, PadSize.getQuantity());
+ if (CGM.shouldZeroInitPadding()) {
+ return llvm::Constant::getNullValue(Ty);
+ }
+ return llvm::UndefValue::get(Ty);
+}
+
struct ConstantAggregateBuilderUtils {
CodeGenModule &CGM;
@@ -61,10 +71,7 @@ struct ConstantAggregateBuilderUtils {
}
llvm::Constant *getPadding(CharUnits PadSize) const {
- llvm::Type *Ty = CGM.CharTy;
- if (PadSize > CharUnits::One())
- Ty = llvm::ArrayType::get(Ty, PadSize.getQuantity());
- return llvm::UndefValue::get(Ty);
+ return ::getPadding(CGM, PadSize);
}
llvm::Constant *getZeroes(CharUnits ZeroSize) const {
@@ -591,6 +598,11 @@ class ConstStructBuilder {
bool Build(const InitListExpr *ILE, bool AllowOverwrite);
bool Build(const APValue &Val, const RecordDecl *RD, bool IsPrimaryBase,
const CXXRecordDecl *VTableClass, CharUnits BaseOffset);
+ bool DoZeroInitPadding(const ASTRecordLayout &Layout, unsigned FieldNo,
+ const FieldDecl &Field, bool AllowOverwrite,
+ CharUnits &SizeSoFar, bool &ZeroFieldSize);
+ bool DoZeroInitPadding(const ASTRecordLayout &Layout, bool AllowOverwrite,
+ CharUnits SizeSoFar);
llvm::Constant *Finalize(QualType Ty);
};
@@ -715,6 +727,10 @@ bool ConstStructBuilder::Build(const InitListExpr *ILE, bool AllowOverwrite) {
if (CXXRD->getNumBases())
return false;
+ const bool ZeroInitPadding = CGM.shouldZeroInitPadding();
+ bool ZeroFieldSize = false;
+ CharUnits SizeSoFar = CharUnits::Zero();
+
for (FieldDecl *Field : RD->fields()) {
++FieldNo;
@@ -732,8 +748,13 @@ bool ConstStructBuilder::Build(const InitListExpr *ILE, bool AllowOverwrite) {
const Expr *Init = nullptr;
if (ElementNo < ILE->getNumInits())
Init = ILE->getInit(ElementNo++);
- if (isa_and_nonnull<NoInitExpr>(Init))
+ if (isa_and_nonnull<NoInitExpr>(Init)) {
+ if (ZeroInitPadding &&
+ !DoZeroInitPadding(Layout, FieldNo, *Field, AllowOverwrite, SizeSoFar,
+ ZeroFieldSize))
+ return false;
continue;
+ }
// Zero-sized fields are not emitted, but their initializers may still
// prevent emission of this struct as a constant.
@@ -743,6 +764,11 @@ bool ConstStructBuilder::Build(const InitListExpr *ILE, bool AllowOverwrite) {
continue;
}
+ if (ZeroInitPadding &&
+ !DoZeroInitPadding(Layout, FieldNo, *Field, AllowOverwrite, SizeSoFar,
+ ZeroFieldSize))
+ return false;
+
// When emitting a DesignatedInitUpdateExpr, a nested InitListExpr
// represents additional overwriting of our current constant value, and not
// a new constant to emit independently.
@@ -768,6 +794,10 @@ bool ConstStructBuilder::Build(const InitListExpr *ILE, bool AllowOverwrite) {
if (!EltInit)
return false;
+ if (ZeroInitPadding && ZeroFieldSize)
+ SizeSoFar += CharUnits::fromQuantity(
+ CGM.getDataLayout().getTypeAllocSize(EltInit->getType()));
+
if (!Field->isBitField()) {
// Handle non-bitfield members.
if (!AppendField(Field, Layout.getFieldOffset(FieldNo), EltInit,
@@ -785,6 +815,9 @@ bool ConstStructBuilder::Build(const InitListExpr *ILE, bool AllowOverwrite) {
}
}
+ if (ZeroInitPadding && !DoZeroInitPadding(Layout, AllowOverwrite, SizeSoFar))
+ return false;
+
return true;
}
@@ -849,6 +882,9 @@ bool ConstStructBuilder::Build(const APValue &Val, const RecordDecl *RD,
unsigned FieldNo = 0;
uint64_t OffsetBits = CGM.getContext().toBits(Offset);
+ const bool ZeroInitPadding = CGM.shouldZeroInitPadding();
+ bool ZeroFieldSize = false;
+ CharUnits SizeSoFar = CharUnits::Zero();
bool AllowOverwrite = false;
for (RecordDecl::field_iterator Field = RD->field_begin(),
@@ -870,6 +906,15 @@ bool ConstStructBuilder::Build(const APValue &Val, const RecordDecl *RD,
if (!EltInit)
return false;
+ if (ZeroInitPadding) {
+ if (!DoZeroInitPadding(Layout, FieldNo, **Field, AllowOverwrite,
+ SizeSoFar, ZeroFieldSize))
+ return false;
+ if (ZeroFieldSize)
+ SizeSoFar += CharUnits::fromQuantity(
+ CGM.getDataLayout().getTypeAllocSize(EltInit->getType()));
+ }
+
if (!Field->isBitField()) {
// Handle non-bitfield members.
if (!AppendField(*Field, Layout.getFieldOffset(FieldNo) + OffsetBits,
@@ -886,7 +931,49 @@ bool ConstStructBuilder::Build(const APValue &Val, const RecordDecl *RD,
return false;
}
}
+ if (ZeroInitPadding && !DoZeroInitPadding(Layout, AllowOverwrite, SizeSoFar))
+ return false;
+
+ return true;
+}
+
+bool ConstStructBuilder::DoZeroInitPadding(
+ const ASTRecordLayout &Layout, unsigned FieldNo, const FieldDecl &Field,
+ bool AllowOverwrite, CharUnits &SizeSoFar, bool &ZeroFieldSize) {
+ uint64_t StartBitOffset = Layout.getFieldOffset(FieldNo);
+ CharUnits StartOffset = CGM.getContext().toCharUnitsFromBits(StartBitOffset);
+ if (SizeSoFar < StartOffset)
+ if (!AppendBytes(SizeSoFar, getPadding(CGM, StartOffset - SizeSoFar),
+ AllowOverwrite))
+ return false;
+
+ if (!Field.isBitField()) {
+ CharUnits FieldSize = CGM.getContext().getTypeSizeInChars(Field.getType());
+ SizeSoFar = StartOffset + FieldSize;
+ ZeroFieldSize = FieldSize.isZero();
+ } else {
+ const CGRecordLayout &RL =
+ CGM.getTypes().getCGRecordLayout(Field.getParent());
+ const CGBitFieldInfo &Info = RL.getBitFieldInfo(&Field);
+ uint64_t EndBitOffset = StartBitOffset + Info.Size;
+ SizeSoFar = CGM.getContext().toCharUnitsFromBits(EndBitOffset);
+ if (EndBitOffset % CGM.getContext().getCharWidth() != 0) {
+ SizeSoFar++;
+ }
+ ZeroFieldSize = Info.Size == 0;
+ }
+ return true;
+}
+bool ConstStructBuilder::DoZeroInitPadding(const ASTRecordLayout &Layout,
+ bool AllowOverwrite,
+ CharUnits SizeSoFar) {
+ CharUnits TotalSize = Layout.getSize();
+ if (SizeSoFar < TotalSize)
+ if (!AppendBytes(SizeSoFar, getPadding(CGM, TotalSize - SizeSoFar),
+ AllowOverwrite))
+ return false;
+ SizeSoFar = TotalSize;
return true;
}
@@ -1127,12 +1214,10 @@ class ConstExprEmitter
assert(CurSize <= TotalSize && "Union size mismatch!");
if (unsigned NumPadBytes = TotalSize - CurSize) {
- llvm::Type *Ty = CGM.CharTy;
- if (NumPadBytes > 1)
- Ty = llvm::ArrayType::get(Ty, NumPadBytes);
-
- Elts.push_back(llvm::UndefValue::get(Ty));
- Types.push_back(Ty);
+ llvm::Constant *Padding =
+ getPadding(CGM, CharUnits::fromQuantity(NumPadBytes));
+ Elts.push_back(Padding);
+ Types.push_back(Padding->getType());
}
llvm::StructType *STy = llvm::StructType::get(VMContext, Types, false);
diff --git a/clang/lib/CodeGen/CodeGenModule.h b/clang/lib/CodeGen/CodeGenModule.h
index c58bb88035ca8a..fcdfef03088da5 100644
--- a/clang/lib/CodeGen/CodeGenModule.h
+++ b/clang/lib/CodeGen/CodeGenModule.h
@@ -1676,6 +1676,57 @@ class CodeGenModule : public CodeGenTypeCache {
MustTailCallUndefinedGlobals.insert(Global);
}
+ bool shouldZeroInitPadding() const {
+ // In C23 (N3096) $6.7.10:
+ // """
+ // If any object is initialized with an empty iniitializer, then it is
+ // subject to default initialization:
+ // - if it is an aggregate, every member is initialized (recursively)
+ // according to these rules, and any padding is initialized to zero bits;
+ // - if it is a union, the first named member is initialized (recursively)
+ // according to these rules, and any padding is initialized to zero bits.
+ //
+ // If the aggregate or union contains elements or members that are
+ // aggregates or unions, these rules apply recursively to the subaggregates
+ // or contained unions.
+ //
+ // If there are fewer initializers in a brace-enclosed list than there are
+ // elements or members of an aggregate, or fewer characters in a string
+ // literal used to initialize an array of known size than there are elements
+ // in the array, the remainder of the aggregate is subject to default
+ // initialization.
+ // """
+ //
+ // From my understanding, the standard is ambiguous in the following two
+ // areas:
+ // 1. For a union type with empty initializer, if the first named member is
+ // not the largest member, then the bytes comes after the first named member
+ // but before padding are left unspecified. An example is:
+ // union U { int a; long long b;};
+ // union U u = {}; // The first 4 bytes are 0, but 4-8 bytes are left
+ // unspecified.
+ //
+ // 2. It only mentions padding for empty initializer, but doesn't mention
+ // padding for a non empty initialization list. And if the aggregation or
+ // union contains elements or members that are aggregates or unions, and
+ // some are non empty initializers, while others are empty initiailizers,
+ // the padding initialization is unclear. An example is:
+ // struct S1 { int a; long long b; };
+ // struct S2 { char c; struct S1 s1; };
+ // // The values for paddings between s2.c and s2.s1.a, between s2.s1.a
+ // and s2.s1.b are unclear.
+ // struct S2 s2 = { 'c' };
+ //
+ // Here we choose to zero initiailize left bytes of a union type. Because
+ // projects like the Linux kernel are relying on this behavior. If we don't
+ // explicitly zero initialize them, the undef values can be optimized to
+ // return gabage data. We also choose to zero initialize paddings for
+ // aggregates and unions, no matter they are initialized by empty
+ // initializers or non empty initializers. This can provide a consistent
+ // behavior. So projects like the Linux kernel can rely on it.
+ return !getLangOpts().CPlusPlus;
+ }
+
private:
bool shouldDropDLLAttribute(const Decl *D, const llvm::GlobalValue *GV) const;
diff --git a/clang/test/CodeGen/2008-07-22-bitfield-init-after-zero-len-array.c b/clang/test/CodeGen/2008-07-22-bitfield-init-after-zero-len-array.c
index b72d689659e602..b639734ef5d4b7 100644
--- a/clang/test/CodeGen/2008-07-22-bitfield-init-after-zero-len-array.c
+++ b/clang/test/CodeGen/2008-07-22-bitfield-init-after-zero-len-array.c
@@ -8,4 +8,4 @@ struct et7 {
52,
};
-// CHECK: @yv7 ={{.*}} global %struct.et7 { [0 x float] zeroinitializer, i8 52 }
+// CHECK: @yv7 ={{.*}} global { [0 x float], i8, [3 x i8] } { [0 x float] zeroinitializer, i8 52, [3 x i8] zeroinitializer }
diff --git a/clang/test/CodeGen/2008-08-07-AlignPadding1.c b/clang/test/CodeGen/2008-08-07-AlignPadding1.c
index 17e88ce02659f0..d69cbc22cc1dfb 100644
--- a/clang/test/CodeGen/2008-08-07-AlignPadding1.c
+++ b/clang/test/CodeGen/2008-08-07-AlignPadding1.c
@@ -20,9 +20,9 @@ struct gc_generation {
#define GEN_HEAD(n) (&generations[n].head)
-// The idea is that there are 6 undefs in this structure initializer to cover
+// The idea is that there are 6 zeroinitializers in this structure initializer to cover
// the padding between elements.
-// CHECK: @generations ={{.*}} global [3 x %struct.gc_generation] [%struct.gc_generation { %union._gc_head { %struct.anon { ptr @generations, ptr @generations, i64 0 }, [8 x i8] undef }, i32 700, i32 0, [8 x i8] undef }, %struct.gc_generation { %union._gc_head { %struct.anon { ptr getelementptr (i8, ptr @generations, i64 48), ptr getelementptr (i8, ptr @generations, i64 48), i64 0 }, [8 x i8] undef }, i32 10, i32 0, [8 x i8] undef }, %struct.gc_generation { %union._gc_head { %struct.anon { ptr getelementptr (i8, ptr @generations, i64 96), ptr getelementptr (i8, ptr @generations, i64 96), i64 0 }, [8 x i8] undef }, i32 10, i32 0, [8 x i8] undef }]
+// CHECK: @generations ={{.*}} global [3 x %struct.gc_generation] [%struct.gc_generation { %union._gc_head { %struct.anon { ptr @generations, ptr @generations, i64 0 }, [8 x i8] zeroinitializer }, i32 700, i32 0, [8 x i8] zeroinitializer }, %struct.gc_generation { %union._gc_head { %struct.anon { ptr getelementptr (i8, ptr @generations, i64 48), ptr getelementptr (i8, ptr @generations, i64 48), i64 0 }, [8 x i8] zeroinitializer }, i32 10, i32 0, [8 x i8] zeroinitializer }, %struct.gc_generation { %union._gc_head { %struct.anon { ptr getelementptr (i8, ptr @generations, i64 96), ptr getelementptr (i8, ptr @generations, i64 96), i64 0 }, [8 x i8] zeroinitializer }, i32 10, i32 0, [8 x i8] zeroinitializer }]
/* linked lists of container objects */
struct gc_generation generations[3] = {
/* PyGC_Head, threshold, count */
diff --git a/clang/test/CodeGen/2009-06-14-anonymous-union-init.c b/clang/test/CodeGen/2009-06-14-anonymous-union-init.c
index 13f6357f7966d9..a4375d7868f01d 100644
--- a/clang/test/CodeGen/2009-06-14-anonymous-union-init.c
+++ b/clang/test/CodeGen/2009-06-14-anonymous-union-init.c
@@ -7,7 +7,7 @@ struct sysfs_dirent {
};
struct sysfs_dirent sysfs_root = { {}, 16877 };
-// CHECK: @sysfs_root = {{.*}}global %struct.sysfs_dirent { %union.anon zeroinitializer, i16 16877 }
+// CHECK: @sysfs_root = {{.*}}global { %union.anon, i16, [2 x i8] } { %union.anon zeroinitializer, i16 16877, [2 x i8] zeroinitializer }
struct Foo {
union { struct empty {} x; };
@@ -16,4 +16,4 @@ struct Foo {
struct Foo foo = { {}, 16877 };
// EMPTY: @foo = {{.*}}global %struct.Foo { i16 16877 }
-// EMPTY-MSVC: @foo = {{.*}}global %struct.Foo { [4 x i8] undef, i16 16877 }
+// EMPTY-MSVC: @foo = {{.*}}global %struct.Foo { [4 x i8] zeroinitializer, i16 16877 }
diff --git a/clang/test/CodeGen/64bit-swiftcall.c b/clang/test/CodeGen/64bit-swiftcall.c
index 7af65ccf556a81..7f8aa02d97ce1f 100644
--- a/clang/test/CodeGen/64bit-swiftcall.c
+++ b/clang/test/CodeGen/64bit-swiftcall.c
@@ -14,8 +14,6 @@
// CHECK-DAG: %struct.atomic_padded = type { { %struct.packed, [7 x i8] } }
// CHECK-DAG: %struct.packed = type <{ i64, i8 }>
-//
-// CHECK: [[STRUCT2_RESULT:@.*]] = private {{.*}} constant [[STRUCT2_TYPE:%.*]] { i32 0, i8 0, i8 undef, i8 0, i32 0, i32 0 }
/*****************************************************************************/
/****************************** PARAMETER ABIS *******************************/
@@ -162,8 +160,8 @@ typedef struct {
} struct_2;
TEST(struct_2);
// CHECK-LABEL: define{{.*}} swiftcc { i64, i64 } @return_struct_2() {{.*}}{
-// CHECK: [[RET:%.*]] = alloca [[STRUCT2_TYPE]], align 4
-// CHECK: call void @llvm.memcpy{{.*}}({{.*}}[[RET]], {{.*}}[[STRUCT2_RESULT]]
+// CHECK: [[RET:%.*]] = alloca [[STRUCT2:%.*]], align 4
+// CHECK: call void @llvm.memset
// CHECK: [[GEP0:%.*]] = getelementptr inbounds nuw { i64, i64 }, ptr [[RET]], i32 0, i32 0
// CHECK: [[T0:%.*]] = load i64, ptr [[GEP0]], align 4
// CHECK: [[GEP1:%.*]] = getelementptr inbounds nuw { i64, i64 }, ptr [[RET]], i32 0, i32 1
@@ -173,7 +171,7 @@ TEST(struct_2);
// CHECK: ret { i64, i64 } [[R1]]
// CHECK: }
// CHECK-LABEL: define{{.*}} swiftcc void @take_struct_2(i64 %0, i64 %1) {{.*}}{
-// CHECK: [[V:%.*]] = alloca [[STRUCT:%.*]], align 4
+// CHECK: [[V:%.*]] = alloca [[STRUCT2]], align 4
// CHECK: [[GEP0:%.*]] = getelementptr inbounds nuw { i64, i64 }, ptr [[V]], i32 0, i32 0
// CHECK: store i64 %0, ptr [[GEP0]], align 4
// CHECK: [[GEP1:%.*]] = getelementptr inbounds nuw { i64, i64 }, ptr [[V]], i32 0, i32 1
@@ -181,7 +179,7 @@ TEST(struct_2);
// CHECK: ret void
// CHECK: }
// CHECK-LABEL: define{{.*}} void @test_struct_2() {{.*}} {
-// CHECK: [[TMP:%.*]] = alloca [[STRUCT2_TYPE]], align 4
+// CHECK: [[TMP:%.*]] = alloca [[STRUCT2]], align 4
// CHECK: [[CALL:%.*]] = call swiftcc { i64, i64 } @return_struct_2()
// CHECK: [[GEP:%.*]] = getelementptr inbounds nuw {{.*}} [[TMP]], i32 0, i32 0
// CHECK: [[T0:%.*]] = extractvalue { i64, i64 } [[CALL]], 0
@@ -254,7 +252,7 @@ typedef union {
TEST(union_het_fp)
// CHECK-LABEL: define{{.*}} swiftcc i64 @return_union_het_fp()
// CHECK: [[RET:%.*]] = alloca [[UNION:%.*]], align 8
-// CHECK: call void @llvm.memcpy{{.*}}(ptr align 8 [[RET]]
+// CHECK: call void @llvm.memset{{.*}}(ptr align 8 [[RET]]
// CHECK: [[GEP:%.*]] = getelementptr inbounds nuw { i64 }, ptr [[RET]], i32 0, i32 0
// CHECK: [[R0:%.*]] = load i64, ptr [[GEP]], align 8
// CHECK: ret i64 [[R0]]
diff --git a/clang/test/CodeGen/arm-swiftcall.c b/clang/test/CodeGen/arm-swiftcall.c
index ec0e3867909a86..677b878c6765da 100644
--- a/clang/test/CodeGen/arm-swiftcall.c
+++ b/clang/test/CodeGen/arm-swiftcall.c
@@ -172,7 +172,7 @@ typedef struct {
TEST(struct_2);
// CHECK-LABEL: define{{.*}} @return_struct_2()
// CHECK: [[RET:%.*]] = alloca [[REC:%.*]], align 4
-// CHECK: @llvm.memcpy
+// CHECK: @llvm.memset
// CHECK: [[T0:%.*]] = getelementptr inbounds nuw [[AGG:{ i32, i32, float, float }]], ptr [[RET]], i32 0, i32 0
// CHECK: [[FIRST:%.*]] = load i32, ptr [[T0]], align 4
// CHECK: [[T0:%.*]] = getelementptr inbounds nuw [[AGG]], ptr [[RET]], i32 0, i32 1
@@ -274,7 +274,7 @@ typedef union {
TEST(union_het_fp)
// CHECK-LABEL: define{{.*}} @return_union_het_fp()
// CHECK: [[RET:%.*]] = alloca [[REC:%.*]], align {{(4|8)}}
-// CHECK: @llvm.memcpy
+// CHECK: @llvm.memset
// CHECK: [[T0:%.*]] = getelementptr inbounds nuw [[AGG:{ i32, i32 }]], ptr [[RET]], i32 0, i32 0
// CHECK: [[FIRST:%.*]] = load i32, ptr [[T0]], align {{(4|8)}}
// CHECK: [[T0:%.*]] = getelementptr inbounds nuw [[AGG]], ptr [[RET]], i32 0, i32 1
diff --git a/clang/test/CodeGen/const-init.c b/clang/test/CodeGen/const-init.c
index ad3e9551199ac2..fc973cb983a80a 100644
--- a/clang/test/CodeGen/const-init.c
+++ b/clang/test/CodeGen/const-init.c
@@ -170,7 +170,7 @@ void g30(void) {
int : 1;
int x;
} a = {};
- // CHECK: @g30.a = internal global %struct.anon.1 <{ i8 undef, i32 0 }>, align 1
+ // CHECK: @g30.a = internal global %struct.anon.1 zeroinitializer, align 1
#pragma pack()
}
@@ -182,7 +182,7 @@ void g31(void) {
short z;
} a = {23122, -12312731, -312};
#pragma pack()
- // CHECK: @g31.a = internal global %struct.anon.2 { i16 23122, i32 -12312731, i16 -312 }, align 4
+ // CHECK: @g31.a = internal global { i16, [2 x i8], i32, i16, [2 x i8] } { i16 23122, [2 x i8] zeroinitializer, i32 -12312731, i16 -312, [2 x i8] zeroinitializer }, align 4
}
// Clang should evaluate this in constant context, so floating point mode should
diff --git a/clang/test/CodeGen/decl.c b/clang/test/CodeGen/decl.c
index a63846b3223da4..97446781fdbd2b 100644
--- a/clang/test/CodeGen/decl.c
+++ b/clang/test/CodeGen/decl.c
@@ -2,10 +2,10 @@
// CHECK: @test1.x = internal constant [12 x i32] [i32 1
// CHECK: @__const.test2.x = private unnamed_addr constant [13 x i32] [i32 1,
-// CHECK: @test5w = {{(dso_local )?}}global { i32, [4 x i8] } { i32 2, [4 x i8] undef }
+// CHECK: @test5w = {{(dso_local )?}}global { i32, [4 x i8] } { i32 2, [4 x i8] zeroinitializer }
// CHECK: @test5y = {{(dso_local )?}}global { double } { double 7.300000e+0{{[0]*}}1 }
-// CHECK: @__const.test6.x = private unnamed_addr constant %struct.SelectDest { i8 1, i8 2, i32 3, i32 0 }
+// CHECK: @__const.test6.x = private unnamed_addr constant { i8, i8, [2 x i8], i32, i32 } { i8 1, i8 2, [2 x i8] zeroinitializer, i32 3, i32 0 }
// CHECK: @test7 = {{(dso_local )?}}global [2 x %struct.test7s] [%struct.test7s { i32 1, i32 2 }, %struct.test7s { i32 4, i32 0 }]
diff --git a/clang/test/CodeGen/designated-initializers.c b/clang/test/CodeGen/designated-initializers.c
index 620b1b90d25758..ac7860db43be77 100644
--- a/clang/test/CodeGen/designated-initializers.c
+++ b/clang/test/CodeGen/designated-initializers.c
@@ -8,7 +8,7 @@ struct foo {
// CHECK: @u ={{.*}} global %union.anon zeroinitializer
union { int i; float f; } u = { };
-// CHECK: @u2 ={{.*}} global { i32, [4 x i8] } { i32 0, [4 x i8] undef }
+// CHECK: @u2 ={{.*}} global { i32, [4 x i8] } zeroinitializer
union { int i; double f; } u2 = { };
// CHECK: @u3 ={{.*}} global %union.anon.1 zeroinitializer
@@ -62,22 +62,22 @@ struct overwrite_string_struct2 {
char L[6];
int M;
} overwrite_string2[] = { { { "foo" }, 1 }, [0].L[2] = 'x'};
-// CHECK: [6 x i8] c"fox\00\00\00", i32 1
+// CHECK: [6 x i8] c"fox\00\00\00", [2 x i8] zeroinitializer, i32 1
struct overwrite_string_struct3 {
char L[3];
int M;
} overwrite_string3[] = { { { "foo" }, 1 }, [0].L[2] = 'x'};
-// CHECK: [3 x i8] c"fox", i32 1
+// CHECK: [3 x i8] c"fox", i8 0, i32 1
struct overwrite_string_struct4 {
char L[3];
int M;
} overwrite_string4[] = { { { "foobar" }, 1 }, [0].L[2] = 'x'};
-// CHECK: [3 x i8] c"fox", i32 1
+// CHECK: [3 x i8] c"fox", i8 0, i32 1
struct overwrite_string_struct5 {
char L[6];
int M;
} overwrite_string5[] = { { { "foo" }, 1 }, [0].L[4] = 'y'};
-// CHECK: [6 x i8] c"foo\00y\00", i32 1
+// CHECK: [6 x i8] c"foo\00y\00", [2 x i8] zeroinitializer, i32 1
// CHECK: @u1 = {{.*}} { i32 65535 }
@@ -138,7 +138,7 @@ union_16644_t union_16644_instance_4[2] =
[1].b[1] = 4
};
-// CHECK: @lab ={{.*}} global { [4 x i8], i32 } { [4 x i8] undef, i32 123 }
+// CHECK: @lab ={{.*}} global { [4 x i8], i32 } { [4 x i8] zeroinitializer, i32 123 }
struct leading_anon_bitfield { int : 32; int n; } lab = { .n = 123 };
struct Base {
diff --git a/clang/test/CodeGen/ext-int.c b/clang/test/CodeGen/ext-int.c
index e3d609a4ba4a2e..aebacd6f22ffc4 100644
--- a/clang/test/CodeGen/ext-int.c
+++ b/clang/test/CodeGen/ext-int.c
@@ -16,7 +16,7 @@
unsigned _BitInt(1) GlobSize1 = 0;
// CHECK: @GlobSize1 = {{.*}}global i8 0
-// CHECK64: @__const.foo.A = private unnamed_addr constant { i32, [4 x i8], <{ i8, [23 x i8] }> } { i32 1, [4 x i8] undef, <{ i8, [23 x i8] }> <{ i8 -86, [23 x i8] zeroinitializer }> }, align 8
+// CHECK64: @__const.foo.A = private unnamed_addr constant { i32, [4 x i8], <{ i8, [23 x i8] }> } { i32 1, [4 x i8] zeroinitializer, <{ i8, [23 x i8] }> <{ i8 -86, [23 x i8] zeroinitializer }> }, align 8
// @BigGlob = global [40 x i8] c"\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF\FF", align 8
// CHECK64: @f.p = internal global <{ i8, i8, [22 x i8] }> <{ i8 16, i8 39, [22 x i8] zeroinitializer }>, align 8
@@ -91,8 +91,8 @@ int foo(int a) {
// CHECK64: %B2 = getelementptr inbounds nuw %struct.S1, ptr %B, i32 0, i32 2
// WIN32: %B2 = getelementptr inbounds nuw %struct.S1, ptr %B, i32 0, i32 2
// LIN32: %B2 = getelementptr inbounds nuw %struct.S1, ptr %B, i32 0, i32 1
- // CHECK: %0 = load i32, ptr %a.addr, align 4
- // CHECK: %conv = sext i32 %0 to i129
+ // CHECK: %[[V1:.+]] = load i32, ptr %a.addr, align 4
+ // CHECK: %conv = sext i32 %[[V1]] to i129
// CHECK64: storedv = sext i129 %conv to i192
// WIN32: storedv = sext i129 %conv to i192
// LIN32: storedv = sext i129 %conv to i160
@@ -102,12 +102,12 @@ int foo(int a) {
// CHECK64: %B3 = getelementptr inbounds nuw %struct.S1, ptr %A, i32 0, i32 2
// WIN32: %B3 = getelementptr inbounds nuw %struct.S1, ptr %A, i32 0, i32 2
// LIN32: %B3 = getelementptr inbounds nuw %struct.S1, ptr %A, i32 0, i32 1
- // CHECK64: %1 = load i192, ptr %B3, align 8
- // WIN32: %1 = load i192, ptr %B3, align 8
- // LIN32: %1 = load i160, ptr %B3, align 4
- // CHECK64: %loadedv = trunc i192 %1 to i129
- // WIN32: %loadedv = trunc i192 %1 to i129
- // LIN32: %loadedv = trunc i160 %1 to i129
+ // CHECK64: %[[V2:.+]] = load i192, ptr %B3, align 8
+ // WIN32: %[[V2:.+]] = load i192, ptr %B3, align 8
+ // LIN32: %[[V2:.+]] = load i160, ptr %B3, align 4
+ // CHECK64: %loadedv = trunc i192 %[[V2]] to i129
+ // WIN32: %loadedv = trunc i192 %[[V2]] to i129
+ // LIN32: %loadedv = trunc i160 %[[V2]] to i129
// CHECK: %conv4 = trunc i129 %loadedv to i32
struct S1 A = {1, 170};
struct S1 B = {1, a};
diff --git a/clang/test/CodeGen/flexible-array-init.c b/clang/test/CodeGen/flexible-array-init.c
index 15a30c15ac966e..17b520fe830942 100644
--- a/clang/test/CodeGen/flexible-array-init.c
+++ b/clang/test/CodeGen/flexible-array-init.c
@@ -14,11 +14,11 @@ struct { int y[]; } b1 = { { 14, 16 } };
// sizeof(c) == 8, so this global should be at least 8 bytes.
struct { int x; char c; char y[]; } c = { 1, 2, { 13, 15 } };
-// CHECK: @c ={{.*}} global { i32, i8, [2 x i8] } { i32 1, i8 2, [2 x i8] c"\0D\0F" }
+// CHECK: @c ={{.*}} global { i32, i8, [2 x i8], i8 } { i32 1, i8 2, [2 x i8] c"\0D\0F", i8 0 }
// sizeof(d) == 8, so this global should be at least 8 bytes.
struct __attribute((packed, aligned(4))) { char a; int x; char z[]; } d = { 1, 2, { 13, 15 } };
-// CHECK: @d ={{.*}} <{ i8, i32, [2 x i8], i8 }> <{ i8 1, i32 2, [2 x i8] c"\0D\0F", i8 undef }>,
+// CHECK: @d ={{.*}} <{ i8, i32, [2 x i8], i8 }> <{ i8 1, i32 2, [2 x i8] c"\0D\0F", i8 0 }>,
// This global needs 9 bytes to hold all the flexible array members.
struct __attribute((packed, aligned(4))) { char a; int x; char z[]; } e = { 1, 2, { 13, 15, 17, 19 } };
@@ -55,21 +55,21 @@ struct { int a; union { int b; short x[]; }; int c; int d; } hf = {1, 2, {}, 3};
// First member is the potential flexible array, initialization requires braces.
struct { int a; union { short x; int b; }; int c; int d; } i = {1, 2, {}, 3};
-// CHECK: @i = global { i32, { i16, [2 x i8] }, i32, i32 } { i32 1, { i16, [2 x i8] } { i16 2, [2 x i8] undef }, i32 0, i32 3 }
+// CHECK: @i = global { i32, { i16, [2 x i8] }, i32, i32 } { i32 1, { i16, [2 x i8] } { i16 2, [2 x i8] zeroinitializer }, i32 0, i32 3 }
struct { int a; union { short x[0]; int b; }; int c; int d; } i0 = {1, {}, 2, 3};
-// CHECK: @i0 = global { i32, { [0 x i16], [4 x i8] }, i32, i32 } { i32 1, { [0 x i16], [4 x i8] } { [0 x i16] zeroinitializer, [4 x i8] undef }, i32 2, i32 3 }
+// CHECK: @i0 = global { i32, { [0 x i16], [4 x i8] }, i32, i32 } { i32 1, { [0 x i16], [4 x i8] } zeroinitializer, i32 2, i32 3 }
struct { int a; union { short x[1]; int b; }; int c; int d; } i1 = {1, {2}, {}, 3};
-// CHECK: @i1 = global { i32, { [1 x i16], [2 x i8] }, i32, i32 } { i32 1, { [1 x i16], [2 x i8] } { [1 x i16] [i16 2], [2 x i8] undef }, i32 0, i32 3 }
+// CHECK: @i1 = global { i32, { [1 x i16], [2 x i8] }, i32, i32 } { i32 1, { [1 x i16], [2 x i8] } { [1 x i16] [i16 2], [2 x i8] zeroinitializer }, i32 0, i32 3 }
struct { int a; union { short x[]; int b; }; int c; int d; } i_f = {4, {}, {}, 6};
-// CHECK: @i_f = global { i32, { [0 x i16], [4 x i8] }, i32, i32 } { i32 4, { [0 x i16], [4 x i8] } { [0 x i16] zeroinitializer, [4 x i8] undef }, i32 0, i32 6 }
+// CHECK: @i_f = global { i32, { [0 x i16], [4 x i8] }, i32, i32 } { i32 4, { [0 x i16], [4 x i8] } zeroinitializer, i32 0, i32 6 }
// Named initializers; order doesn't matter.
struct { int a; union { int b; short x; }; int c; int d; } hn = {.a = 1, .x = 2, .c = 3};
-// CHECK: @hn = global { i32, { i16, [2 x i8] }, i32, i32 } { i32 1, { i16, [2 x i8] } { i16 2, [2 x i8] undef }, i32 3, i32 0 }
+// CHECK: @hn = global { i32, { i16, [2 x i8] }, i32, i32 } { i32 1, { i16, [2 x i8] } { i16 2, [2 x i8] zeroinitializer }, i32 3, i32 0 }
struct { int a; union { int b; short x[0]; }; int c; int d; } hn0 = {.a = 1, .x = {2}, .c = 3};
-// CHECK: @hn0 = global { i32, { [0 x i16], [4 x i8] }, i32, i32 } { i32 1, { [0 x i16], [4 x i8] } { [0 x i16] zeroinitializer, [4 x i8] undef }, i32 3, i32 0 }
+// CHECK: @hn0 = global { i32, { [0 x i16], [4 x i8] }, i32, i32 } { i32 1, { [0 x i16], [4 x i8] } zeroinitializer, i32 3, i32 0 }
struct { int a; union { int b; short x[1]; }; int c; int d; } hn1 = {.a = 1, .x = {2}, .c = 3};
-// CHECK: @hn1 = global { i32, { [1 x i16], [2 x i8] }, i32, i32 } { i32 1, { [1 x i16], [2 x i8] } { [1 x i16] [i16 2], [2 x i8] undef }, i32 3, i32 0 }
+// CHECK: @hn1 = global { i32, { [1 x i16], [2 x i8] }, i32, i32 } { i32 1, { [1 x i16], [2 x i8] } { [1 x i16] [i16 2], [2 x i8] zeroinitializer }, i32 3, i32 0 }
struct { char a[]; } empty_struct = {};
// CHECK: @empty_struct ={{.*}} global %struct.anon{{.*}} zeroinitializer, align 1
@@ -96,10 +96,10 @@ union { char a[]; } only_in_union0 = {0};
// CHECK: @only_in_union0 = global { [1 x i8] } zeroinitializer, align 1
union { char a[]; int b; } first_in_union = {};
-// CHECK: @first_in_union = global { [0 x i8], [4 x i8] } { [0 x i8] zeroinitializer, [4 x i8] undef }, align 4
+// CHECK: @first_in_union = global { [0 x i8], [4 x i8] } zeroinitializer, align 4
union { char a[]; int b; } first_in_union0 = {0};
-// CHECK: @first_in_union0 = global { [1 x i8], [3 x i8] } { [1 x i8] zeroinitializer, [3 x i8] undef }, align 4
+// CHECK: @first_in_union0 = global { [1 x i8], [3 x i8] } zeroinitializer, align 4
union { char a[]; int b; } first_in_union123 = { {1, 2, 3} };
-// CHECK: @first_in_union123 = global { [3 x i8], i8 } { [3 x i8] c"\01\02\03", i8 undef }, align 4
+// CHECK: @first_in_union123 = global { [3 x i8], i8 } { [3 x i8] c"\01\02\03", i8 0 }, align 4
diff --git a/clang/test/CodeGen/global-init.c b/clang/test/CodeGen/global-init.c
index 7f1d675b97c09e..b156466dbaaffc 100644
--- a/clang/test/CodeGen/global-init.c
+++ b/clang/test/CodeGen/global-init.c
@@ -33,7 +33,7 @@ struct ManyFields {
int f;
};
-// CHECK: global %struct.ManyFields { i32 1, i32 2, i32 0, i8 0, i32 0, i32 0 }
+// CHECK: global { i32, i32, i32, i8, [3 x i8], i32, i32 } { i32 1, i32 2, i32 0, i8 0, [3 x i8] zeroinitializer, i32 0, i32 0 }
struct ManyFields FewInits = {1, 2};
diff --git a/clang/test/CodeGen/init.c b/clang/test/CodeGen/init.c
index cbf615bb9ddfea..27f427dff8f79e 100644
--- a/clang/test/CodeGen/init.c
+++ b/clang/test/CodeGen/init.c
@@ -187,25 +187,6 @@ void nonzeroMemsetf64(void) {
// CHECK: call void @llvm.memset.p0.i32(ptr {{.*}}, i8 68, i32 56, i1 false)
}
-void nonzeroPaddedUnionMemset(void) {
- union U { char c; int i; };
- union U arr[9] = { 0xF0, 0xF0, 0xF0, 0xF0, 0xF0, 0xF0, 0xF0, 0xF0, 0xF0, };
- // CHECK-LABEL: @nonzeroPaddedUnionMemset(
- // CHECK-NOT: store
- // CHECK-NOT: memcpy
- // CHECK: call void @llvm.memset.p0.i32(ptr {{.*}}, i8 -16, i32 36, i1 false)
-}
-
-void nonzeroNestedMemset(void) {
- union U { char c; int i; };
- struct S { union U u; short i; };
- struct S arr[5] = { { {0xF0}, 0xF0F0 }, { {0xF0}, 0xF0F0 }, { {0xF0}, 0xF0F0 }, { {0xF0}, 0xF0F0 }, { {0xF0}, 0xF0F0 }, };
- // CHECK-LABEL: @nonzeroNestedMemset(
- // CHECK-NOT: store
- // CHECK-NOT: memcpy
- // CHECK: call void @llvm.memset.p0.i32(ptr {{.*}}, i8 -16, i32 40, i1 false)
-}
-
// PR9257
struct test11S {
int A[10];
diff --git a/clang/test/CodeGen/linux-kernel-struct-union-initializer.c b/clang/test/CodeGen/linux-kernel-struct-union-initializer.c
new file mode 100644
index 00000000000000..8c0eb3702cf0b5
--- /dev/null
+++ b/clang/test/CodeGen/linux-kernel-struct-union-initializer.c
@@ -0,0 +1,275 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --check-globals all --version 5
+// RUN: %clang_cc1 -triple x86_64-unknown-unknown -std=gnu11 -verify -emit-llvm %s -o - | FileCheck %s
+// expected-no-diagnostics
+
+union U1 {
+ int x;
+ char y[16];
+};
+
+struct S1 {
+ int x;
+ union U1 y;
+};
+
+union U2 {
+ int x;
+ char y[16];
+} __attribute__((__aligned__(32)));
+
+struct S2 {
+ int x;
+ long long y;
+ char z[8];
+} __attribute__((__aligned__(32)));
+
+struct S3 {
+ char x;
+ unsigned char y : 4;
+ unsigned char z : 7;
+} __attribute__((packed));
+
+union U1 global_u1 = {};
+
+union U1 global_u2 = {3};
+
+union U1 global_u2_from_cast = (union U1)3;
+
+struct S1 global_s1 = {};
+
+struct S1 global_s2 = {
+ .x = 3,
+};
+
+struct S1 global_s3 = {.x = 3, .y = {.x = 6}};
+
+const union U1 global_const_u1 = {4};
+struct S1 global_s3_from_const_u1 = {.y = global_const_u1};
+
+union U2 global_u3 = {};
+
+struct S2 global_s4 = {};
+
+struct S2 global_s5 = {.x = 1};
+
+struct S3 global_s6 = {101, 15, 123};
+
+// Test empty initializer for union.
+//.
+// CHECK: @global_u1 = global %union.U1 zeroinitializer, align 4
+// CHECK: @global_u2 = global %union.U1 { i32 3, [12 x i8] zeroinitializer }, align 4
+// CHECK: @global_u2_from_cast = global { i32, [12 x i8] } { i32 3, [12 x i8] zeroinitializer }, align 4
+// CHECK: @global_s1 = global %struct.S1 zeroinitializer, align 4
+// CHECK: @global_s2 = global %struct.S1 { i32 3, %union.U1 zeroinitializer }, align 4
+// CHECK: @global_s3 = global %struct.S1 { i32 3, %union.U1 { i32 6, [12 x i8] zeroinitializer } }, align 4
+// CHECK: @global_const_u1 = constant %union.U1 { i32 4, [12 x i8] zeroinitializer }, align 4
+// CHECK: @global_s3_from_const_u1 = global %struct.S1 { i32 0, %union.U1 { i32 4, [12 x i8] zeroinitializer } }, align 4
+// CHECK: @global_u3 = global %union.U2 zeroinitializer, align 32
+// CHECK: @global_s4 = global { i32, [4 x i8], i64, [8 x i8], [8 x i8] } zeroinitializer, align 32
+// CHECK: @global_s5 = global { i32, [4 x i8], i64, [8 x i8], [8 x i8] } { i32 1, [4 x i8] zeroinitializer, i64 0, [8 x i8] zeroinitializer, [8 x i8] zeroinitializer }, align 32
+// CHECK: @global_s6 = global { i8, i8, i8 } { i8 101, i8 -65, i8 7 }, align 1
+// CHECK: @test2.a = internal global %union.U1 zeroinitializer, align 4
+// CHECK: @__const.test3.a = private unnamed_addr constant %union.U1 { i32 3, [12 x i8] zeroinitializer }, align 4
+// CHECK: @test4.a = internal global %union.U1 { i32 3, [12 x i8] zeroinitializer }, align 4
+// CHECK: @test6.s = internal global %struct.S1 zeroinitializer, align 4
+// CHECK: @__const.test7.s = private unnamed_addr constant %struct.S1 { i32 3, %union.U1 zeroinitializer }, align 4
+// CHECK: @test8.s = internal global %struct.S1 { i32 3, %union.U1 zeroinitializer }, align 4
+// CHECK: @__const.test9.s = private unnamed_addr constant %struct.S1 { i32 3, %union.U1 { i32 6, [12 x i8] zeroinitializer } }, align 4
+// CHECK: @test10.s = internal global %struct.S1 { i32 3, %union.U1 { i32 6, [12 x i8] zeroinitializer } }, align 4
+// CHECK: @test12.a = internal global %union.U2 zeroinitializer, align 32
+// CHECK: @test14.s = internal global { i32, [4 x i8], i64, [8 x i8], [8 x i8] } zeroinitializer, align 32
+// CHECK: @__const.test15.s = private unnamed_addr constant { i32, [4 x i8], i64, [8 x i8], [8 x i8] } { i32 1, [4 x i8] zeroinitializer, i64 0, [8 x i8] zeroinitializer, [8 x i8] zeroinitializer }, align 32
+// CHECK: @test16.s = internal global { i32, [4 x i8], i64, [8 x i8], [8 x i8] } { i32 1, [4 x i8] zeroinitializer, i64 0, [8 x i8] zeroinitializer, [8 x i8] zeroinitializer }, align 32
+//.
+// CHECK-LABEL: define dso_local void @test1(
+// CHECK-SAME: ) #[[ATTR0:[0-9]+]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[A:%.*]] = alloca [[UNION_U1:%.*]], align 4
+// CHECK-NEXT: call void @llvm.memset.p0.i64(ptr align 4 [[A]], i8 0, i64 16, i1 false)
+// CHECK-NEXT: ret void
+//
+void test1() {
+ union U1 a = {};
+}
+
+// Test empty initializer for union. Use static variable.
+// CHECK-LABEL: define dso_local void @test2(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: ret void
+//
+void test2() {
+ static union U1 a = {};
+}
+
+// Test only initializing a small field for union.
+// CHECK-LABEL: define dso_local void @test3(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[A:%.*]] = alloca [[UNION_U1:%.*]], align 4
+// CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[A]], ptr align 4 @__const.test3.a, i64 16, i1 false)
+// CHECK-NEXT: ret void
+//
+void test3() {
+ union U1 a = {3};
+}
+
+// Test only initializing a small field for union. Use static variable.
+// CHECK-LABEL: define dso_local void @test4(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: ret void
+//
+void test4() {
+ static union U1 a = {3};
+}
+
+// Test union in struct. Use empty initializer for the struct.
+// CHECK-LABEL: define dso_local void @test5(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[S:%.*]] = alloca [[STRUCT_S1:%.*]], align 4
+// CHECK-NEXT: call void @llvm.memset.p0.i64(ptr align 4 [[S]], i8 0, i64 20, i1 false)
+// CHECK-NEXT: ret void
+//
+void test5() {
+ struct S1 s = {};
+}
+
+// Test union in struct. Use empty initializer for the struct. Use static variable.
+// CHECK-LABEL: define dso_local void @test6(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: ret void
+//
+void test6() {
+ static struct S1 s = {};
+}
+
+// Test union in struct. Initialize other fields of the struct.
+// CHECK-LABEL: define dso_local void @test7(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[S:%.*]] = alloca [[STRUCT_S1:%.*]], align 4
+// CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[S]], ptr align 4 @__const.test7.s, i64 20, i1 false)
+// CHECK-NEXT: ret void
+//
+void test7() {
+ struct S1 s = {
+ .x = 3,
+ };
+}
+
+// Test union in struct. Initialize other fields of the struct. Use static variable.
+// CHECK-LABEL: define dso_local void @test8(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: ret void
+//
+void test8() {
+ static struct S1 s = {
+ .x = 3,
+ };
+}
+
+// Test union in struct. Initialize a small field for union.
+// CHECK-LABEL: define dso_local void @test9(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[S:%.*]] = alloca [[STRUCT_S1:%.*]], align 4
+// CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[S]], ptr align 4 @__const.test9.s, i64 20, i1 false)
+// CHECK-NEXT: ret void
+//
+void test9() {
+ struct S1 s = {.x = 3,
+ .y = {
+ .x = 6,
+ }};
+}
+
+// Test union in struct. Initialize a small field for union. Use static variable.
+// CHECK-LABEL: define dso_local void @test10(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: ret void
+//
+void test10() {
+ static struct S1 s = {.x = 3,
+ .y = {
+ .x = 6,
+ }};
+}
+
+// Test empty initializer for union with padding.
+// CHECK-LABEL: define dso_local void @test11(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[A:%.*]] = alloca [[UNION_U2:%.*]], align 32
+// CHECK-NEXT: call void @llvm.memset.p0.i64(ptr align 32 [[A]], i8 0, i64 32, i1 false)
+// CHECK-NEXT: ret void
+//
+void test11() {
+ union U2 a = {};
+}
+
+// Test empty initializer for union with padding. Use static variable.
+// CHECK-LABEL: define dso_local void @test12(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: ret void
+//
+void test12() {
+ static union U2 a = {};
+}
+
+// Test empty initializer for struct with padding.
+// CHECK-LABEL: define dso_local void @test13(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[S:%.*]] = alloca [[STRUCT_S2:%.*]], align 32
+// CHECK-NEXT: call void @llvm.memset.p0.i64(ptr align 32 [[S]], i8 0, i64 32, i1 false)
+// CHECK-NEXT: ret void
+//
+void test13() {
+ struct S2 s = {};
+}
+
+// Test empty initializer for struct with padding. Use static variable.
+// CHECK-LABEL: define dso_local void @test14(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: ret void
+//
+void test14() {
+ static struct S2 s = {};
+}
+
+// Test partial initialization for struct with padding.
+// CHECK-LABEL: define dso_local void @test15(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[S:%.*]] = alloca [[STRUCT_S2:%.*]], align 32
+// CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 32 [[S]], ptr align 32 @__const.test15.s, i64 32, i1 false)
+// CHECK-NEXT: ret void
+//
+void test15() {
+ struct S2 s = {.x = 1};
+}
+
+// Test partial initialization for struct with padding. Use static variable.
+// CHECK-LABEL: define dso_local void @test16(
+// CHECK-SAME: ) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: ret void
+//
+void test16() {
+ static struct S2 s = {.x = 1};
+}
+//.
+// CHECK: attributes #[[ATTR0]] = { noinline nounwind optnone "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-features"="+cx8,+mmx,+sse,+sse2,+x87" }
+// CHECK: attributes #[[ATTR1:[0-9]+]] = { nocallback nofree nounwind willreturn memory(argmem: write) }
+// CHECK: attributes #[[ATTR2:[0-9]+]] = { nocallback nofree nounwind willreturn memory(argmem: readwrite) }
+//.
+// CHECK: [[META0:![0-9]+]] = !{i32 1, !"wchar_size", i32 4}
+// CHECK: [[META1:![0-9]+]] = !{!"{{.*}}clang version {{.*}}"}
+//.
diff --git a/clang/test/CodeGen/linux-kernel-struct-union-initializer2.c b/clang/test/CodeGen/linux-kernel-struct-union-initializer2.c
new file mode 100644
index 00000000000000..bf8154601b4b5e
--- /dev/null
+++ b/clang/test/CodeGen/linux-kernel-struct-union-initializer2.c
@@ -0,0 +1,178 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --check-globals all --version 5
+// RUN: %clang_cc1 -triple x86_64-unknown-unknown -std=gnu11 -verify -emit-llvm %s -o - | FileCheck %s
+// expected-no-diagnostics
+
+union U1 {
+ int x;
+ char y[5];
+};
+
+struct S1 {
+ int x;
+ long long y;
+};
+
+struct S2 {
+ unsigned char b1 : 3; // 1st 3 bits (in 1st byte) are b1
+ unsigned char : 2; // next 2 bits (in 1st byte) are blocked out as unused
+ unsigned char b2 : 6; // 6 bits for b2 - doesn't fit into the 1st byte => starts a 2nd
+ unsigned char b3 : 2; // 2 bits for b3 - next (and final) bits in the 2nd byte
+ int i;
+};
+
+struct S3 {
+ int x;
+} __attribute__((__aligned__(8)));
+
+struct S4 {
+ int a;
+ union U1 b;
+};
+
+struct S5 {
+ char x;
+ unsigned char y : 4;
+ unsigned char z : 7;
+} __attribute__((packed));
+
+// Test non-const initializer for union with padding.
+// CHECK-LABEL: define dso_local void @test1(
+// CHECK-SAME: i32 noundef [[X:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[X_ADDR:%.*]] = alloca i32, align 4
+// CHECK-NEXT: [[A:%.*]] = alloca [[UNION_U1:%.*]], align 4
+// CHECK-NEXT: store i32 [[X]], ptr [[X_ADDR]], align 4
+// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[X_ADDR]], align 4
+// CHECK-NEXT: store i32 [[TMP0]], ptr [[A]], align 4
+// CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[A]], i64 4
+// CHECK-NEXT: call void @llvm.memset.p0.i64(ptr align 4 [[TMP1]], i8 0, i64 4, i1 false)
+// CHECK-NEXT: ret void
+//
+void test1(int x) {
+ union U1 a = {x};
+}
+
+// Test non-const initializer for struct with padding.
+// CHECK-LABEL: define dso_local void @test2(
+// CHECK-SAME: i64 noundef [[Y:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[Y_ADDR:%.*]] = alloca i64, align 8
+// CHECK-NEXT: [[S:%.*]] = alloca [[STRUCT_S1:%.*]], align 8
+// CHECK-NEXT: store i64 [[Y]], ptr [[Y_ADDR]], align 8
+// CHECK-NEXT: [[X:%.*]] = getelementptr inbounds nuw [[STRUCT_S1]], ptr [[S]], i32 0, i32 0
+// CHECK-NEXT: store i32 0, ptr [[X]], align 8
+// CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[S]], i64 4
+// CHECK-NEXT: call void @llvm.memset.p0.i64(ptr align 4 [[TMP0]], i8 0, i64 4, i1 false)
+// CHECK-NEXT: [[Y1:%.*]] = getelementptr inbounds nuw [[STRUCT_S1]], ptr [[S]], i32 0, i32 1
+// CHECK-NEXT: [[TMP1:%.*]] = load i64, ptr [[Y_ADDR]], align 8
+// CHECK-NEXT: store i64 [[TMP1]], ptr [[Y1]], align 8
+// CHECK-NEXT: ret void
+//
+void test2(long long y) {
+ struct S1 s = {.y = y};
+}
+
+// Test non-const initializer for struct with padding and bit fields.
+// CHECK-LABEL: define dso_local void @test3(
+// CHECK-SAME: i8 noundef zeroext [[B:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[B_ADDR:%.*]] = alloca i8, align 1
+// CHECK-NEXT: [[S:%.*]] = alloca [[STRUCT_S2:%.*]], align 4
+// CHECK-NEXT: store i8 [[B]], ptr [[B_ADDR]], align 1
+// CHECK-NEXT: [[TMP0:%.*]] = load i8, ptr [[B_ADDR]], align 1
+// CHECK-NEXT: [[TMP1:%.*]] = zext i8 [[TMP0]] to i16
+// CHECK-NEXT: [[BF_LOAD:%.*]] = load i16, ptr [[S]], align 4
+// CHECK-NEXT: [[BF_VALUE:%.*]] = and i16 [[TMP1]], 7
+// CHECK-NEXT: [[BF_CLEAR:%.*]] = and i16 [[BF_LOAD]], -8
+// CHECK-NEXT: [[BF_SET:%.*]] = or i16 [[BF_CLEAR]], [[BF_VALUE]]
+// CHECK-NEXT: store i16 [[BF_SET]], ptr [[S]], align 4
+// CHECK-NEXT: [[BF_LOAD1:%.*]] = load i16, ptr [[S]], align 4
+// CHECK-NEXT: [[BF_CLEAR2:%.*]] = and i16 [[BF_LOAD1]], -16129
+// CHECK-NEXT: [[BF_SET3:%.*]] = or i16 [[BF_CLEAR2]], 0
+// CHECK-NEXT: store i16 [[BF_SET3]], ptr [[S]], align 4
+// CHECK-NEXT: [[BF_LOAD4:%.*]] = load i16, ptr [[S]], align 4
+// CHECK-NEXT: [[BF_CLEAR5:%.*]] = and i16 [[BF_LOAD4]], 16383
+// CHECK-NEXT: [[BF_SET6:%.*]] = or i16 [[BF_CLEAR5]], 0
+// CHECK-NEXT: store i16 [[BF_SET6]], ptr [[S]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[S]], i64 2
+// CHECK-NEXT: call void @llvm.memset.p0.i64(ptr align 2 [[TMP2]], i8 0, i64 2, i1 false)
+// CHECK-NEXT: [[I:%.*]] = getelementptr inbounds nuw [[STRUCT_S2]], ptr [[S]], i32 0, i32 1
+// CHECK-NEXT: store i32 0, ptr [[I]], align 4
+// CHECK-NEXT: ret void
+//
+void test3(unsigned char b) {
+ struct S2 s = {.b1 = b};
+}
+
+// Test non-const initializer for struct with padding at the end of the struct.
+// CHECK-LABEL: define dso_local void @test4(
+// CHECK-SAME: i32 noundef [[X:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[X_ADDR:%.*]] = alloca i32, align 4
+// CHECK-NEXT: [[S:%.*]] = alloca [[STRUCT_S3:%.*]], align 8
+// CHECK-NEXT: store i32 [[X]], ptr [[X_ADDR]], align 4
+// CHECK-NEXT: [[X1:%.*]] = getelementptr inbounds nuw [[STRUCT_S3]], ptr [[S]], i32 0, i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[X_ADDR]], align 4
+// CHECK-NEXT: store i32 [[TMP0]], ptr [[X1]], align 8
+// CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[S]], i64 4
+// CHECK-NEXT: call void @llvm.memset.p0.i64(ptr align 4 [[TMP1]], i8 0, i64 4, i1 false)
+// CHECK-NEXT: ret void
+//
+void test4(int x) {
+ struct S3 s = {x};
+}
+
+// Test non-const initializer for union in struct.
+// CHECK-LABEL: define dso_local void @test5(
+// CHECK-SAME: i32 noundef [[A:%.*]], i32 noundef [[B:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
+// CHECK-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
+// CHECK-NEXT: [[S:%.*]] = alloca [[STRUCT_S4:%.*]], align 4
+// CHECK-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4
+// CHECK-NEXT: store i32 [[B]], ptr [[B_ADDR]], align 4
+// CHECK-NEXT: [[A1:%.*]] = getelementptr inbounds nuw [[STRUCT_S4]], ptr [[S]], i32 0, i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4
+// CHECK-NEXT: store i32 [[TMP0]], ptr [[A1]], align 4
+// CHECK-NEXT: [[B2:%.*]] = getelementptr inbounds nuw [[STRUCT_S4]], ptr [[S]], i32 0, i32 1
+// CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[B_ADDR]], align 4
+// CHECK-NEXT: store i32 [[TMP1]], ptr [[B2]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[B2]], i64 4
+// CHECK-NEXT: call void @llvm.memset.p0.i64(ptr align 4 [[TMP2]], i8 0, i64 4, i1 false)
+// CHECK-NEXT: ret void
+//
+void test5(int a, int b) {
+ struct S4 s = {a, {b}};
+}
+
+// CHECK-LABEL: define dso_local void @test6(
+// CHECK-SAME: i8 noundef signext [[X:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT: [[ENTRY:.*:]]
+// CHECK-NEXT: [[X_ADDR:%.*]] = alloca i8, align 1
+// CHECK-NEXT: [[S:%.*]] = alloca [[STRUCT_S5:%.*]], align 1
+// CHECK-NEXT: store i8 [[X]], ptr [[X_ADDR]], align 1
+// CHECK-NEXT: [[X1:%.*]] = getelementptr inbounds nuw [[STRUCT_S5]], ptr [[S]], i32 0, i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = load i8, ptr [[X_ADDR]], align 1
+// CHECK-NEXT: store i8 [[TMP0]], ptr [[X1]], align 1
+// CHECK-NEXT: [[Y:%.*]] = getelementptr inbounds nuw [[STRUCT_S5]], ptr [[S]], i32 0, i32 1
+// CHECK-NEXT: [[BF_LOAD:%.*]] = load i16, ptr [[Y]], align 1
+// CHECK-NEXT: [[BF_CLEAR:%.*]] = and i16 [[BF_LOAD]], -16
+// CHECK-NEXT: [[BF_SET:%.*]] = or i16 [[BF_CLEAR]], 0
+// CHECK-NEXT: store i16 [[BF_SET]], ptr [[Y]], align 1
+// CHECK-NEXT: [[Z:%.*]] = getelementptr inbounds nuw [[STRUCT_S5]], ptr [[S]], i32 0, i32 1
+// CHECK-NEXT: [[BF_LOAD2:%.*]] = load i16, ptr [[Z]], align 1
+// CHECK-NEXT: [[BF_CLEAR3:%.*]] = and i16 [[BF_LOAD2]], -2033
+// CHECK-NEXT: [[BF_SET4:%.*]] = or i16 [[BF_CLEAR3]], 0
+// CHECK-NEXT: store i16 [[BF_SET4]], ptr [[Z]], align 1
+// CHECK-NEXT: ret void
+//
+void test6(char x) {
+ struct S5 s = {.x = x};
+}
+//.
+// CHECK: attributes #[[ATTR0]] = { noinline nounwind optnone "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-features"="+cx8,+mmx,+sse,+sse2,+x87" }
+// CHECK: attributes #[[ATTR1:[0-9]+]] = { nocallback nofree nounwind willreturn memory(argmem: write) }
+//.
+// CHECK: [[META0:![0-9]+]] = !{i32 1, !"wchar_size", i32 4}
+// CHECK: [[META1:![0-9]+]] = !{!"{{.*}}clang version {{.*}}"}
+//.
diff --git a/clang/test/CodeGen/mingw-long-double.c b/clang/test/CodeGen/mingw-long-double.c
index 4be97526f96319..0fc8f015096827 100644
--- a/clang/test/CodeGen/mingw-long-double.c
+++ b/clang/test/CodeGen/mingw-long-double.c
@@ -11,12 +11,9 @@ struct {
char c;
long double ldb;
} agggregate_LD = {};
-// GNU32: %struct.anon = type { i8, x86_fp80 }
-// GNU32: @agggregate_LD = dso_local global %struct.anon zeroinitializer, align 4
-// GNU64: %struct.anon = type { i8, x86_fp80 }
-// GNU64: @agggregate_LD = dso_local global %struct.anon zeroinitializer, align 16
-// MSC64: %struct.anon = type { i8, double }
-// MSC64: @agggregate_LD = dso_local global %struct.anon zeroinitializer, align 8
+// GNU32: @agggregate_LD = dso_local global { i8, [3 x i8], x86_fp80 } zeroinitializer, align 4
+// GNU64: @agggregate_LD = dso_local global { i8, [15 x i8], x86_fp80 } zeroinitializer, align 16
+// MSC64: @agggregate_LD = dso_local global { i8, [7 x i8], double } zeroinitializer, align 8
long double dataLD = 1.0L;
// GNU32: @dataLD = dso_local global x86_fp80 0xK3FFF8000000000000000, align 4
diff --git a/clang/test/CodeGen/mms-bitfields.c b/clang/test/CodeGen/mms-bitfields.c
index 49c5c1c3e7d40d..2ccce326c7131d 100644
--- a/clang/test/CodeGen/mms-bitfields.c
+++ b/clang/test/CodeGen/mms-bitfields.c
@@ -61,5 +61,5 @@ union HEADER {
struct Inner variable = { 1,0,1, 21 };
union HEADER hdr = {{1,2,3,4}};
-// CHECK: @variable ={{.*}} global { i8, [3 x i8], i8, i8, i8, i8 } { i8 5, [3 x i8] undef, i8 21, i8 0, i8 0, i8 0 }, align 1
-// CHECK: @hdr ={{.*}} global { { i8, i8, [2 x i8], i8, i8, i8, i8, i8, [3 x i8] } } { { i8, i8, [2 x i8], i8, i8, i8, i8, i8, [3 x i8] } { i8 8, i8 0, [2 x i8] undef, i8 2, i8 0, i8 0, i8 3, i8 4, [3 x i8] undef } }, align 1
+// CHECK: @variable ={{.*}} global { i8, [3 x i8], i8, i8, i8, i8 } { i8 5, [3 x i8] zeroinitializer, i8 21, i8 0, i8 0, i8 0 }, align 1
+// CHECK: @hdr ={{.*}} global { { i8, i8, [2 x i8], i8, i8, i8, i8, i8, [3 x i8] } } { { i8, i8, [2 x i8], i8, i8, i8, i8, i8, [3 x i8] } { i8 8, i8 0, [2 x i8] zeroinitializer, i8 2, i8 0, i8 0, i8 3, i8 4, [3 x i8] zeroinitializer } }, align 1
diff --git a/clang/test/CodeGen/union-init2.c b/clang/test/CodeGen/union-init2.c
index 048ff00517b4e8..ee35e78a4f3010 100644
--- a/clang/test/CodeGen/union-init2.c
+++ b/clang/test/CodeGen/union-init2.c
@@ -2,11 +2,11 @@
// RUN: %clang_cc1 -x c++ %s -emit-llvm -triple x86_64-linux-gnu -o - | FileCheck %s --check-prefixes=CHECK-CXX
// Make sure we generate something sane instead of a ptrtoint
-// CHECK: @r, [4 x i8] undef
+// CHECK: @r, [4 x i8] zeroinitializer
union x {long long b;union x* a;} r = {.a = &r};
-// CHECK: global { [3 x i8], [5 x i8] } { [3 x i8] zeroinitializer, [5 x i8] undef }
+// CHECK: global { [3 x i8], [5 x i8] } zeroinitializer
union z {
char a[3];
long long b;
diff --git a/clang/test/CodeGen/windows-swiftcall.c b/clang/test/CodeGen/windows-swiftcall.c
index bc7832d9d3ac28..41569c2606622f 100644
--- a/clang/test/CodeGen/windows-swiftcall.c
+++ b/clang/test/CodeGen/windows-swiftcall.c
@@ -5,8 +5,6 @@
#define ERROR __attribute__((swift_error_result))
#define CONTEXT __attribute__((swift_context))
-// CHECK: [[STRUCT2_RESULT:@.*]] = private {{.*}} constant [[STRUCT2_TYPE:%.*]] { i32 0, i8 0, i8 undef, i8 0, i32 0, i32 0 }
-
/*****************************************************************************/
/****************************** PARAMETER ABIS *******************************/
/*****************************************************************************/
@@ -142,8 +140,8 @@ typedef struct {
} struct_2;
TEST(struct_2);
// CHECK-LABEL: define dso_local swiftcc { i64, i64 } @return_struct_2() {{.*}}{
-// CHECK: [[RET:%.*]] = alloca [[STRUCT2_TYPE]], align 4
-// CHECK: call void @llvm.memcpy{{.*}}({{.*}}[[RET]], {{.*}}[[STRUCT2_RESULT]]
+// CHECK: [[RET:%.*]] = alloca [[STRUCT2:%.*]], align 4
+// CHECK: call void @llvm.memset
// CHECK: [[GEP0:%.*]] = getelementptr inbounds nuw { i64, i64 }, ptr [[RET]], i32 0, i32 0
// CHECK: [[T0:%.*]] = load i64, ptr [[GEP0]], align 4
// CHECK: [[GEP1:%.*]] = getelementptr inbounds nuw { i64, i64 }, ptr [[RET]], i32 0, i32 1
@@ -153,7 +151,7 @@ TEST(struct_2);
// CHECK: ret { i64, i64 } [[R1]]
// CHECK: }
// CHECK-LABEL: define dso_local swiftcc void @take_struct_2(i64 %0, i64 %1) {{.*}}{
-// CHECK: [[V:%.*]] = alloca [[STRUCT:%.*]], align 4
+// CHECK: [[V:%.*]] = alloca [[STRUCT2]], align 4
// CHECK: [[GEP0:%.*]] = getelementptr inbounds nuw { i64, i64 }, ptr [[V]], i32 0, i32 0
// CHECK: store i64 %0, ptr [[GEP0]], align 4
// CHECK: [[GEP1:%.*]] = getelementptr inbounds nuw { i64, i64 }, ptr [[V]], i32 0, i32 1
@@ -161,7 +159,7 @@ TEST(struct_2);
// CHECK: ret void
// CHECK: }
// CHECK-LABEL: define dso_local void @test_struct_2() {{.*}} {
-// CHECK: [[TMP:%.*]] = alloca [[STRUCT2_TYPE]], align 4
+// CHECK: [[TMP:%.*]] = alloca [[STRUCT2]], align 4
// CHECK: [[CALL:%.*]] = call swiftcc { i64, i64 } @return_struct_2()
// CHECK: [[GEP:%.*]] = getelementptr inbounds nuw {{.*}} [[TMP]], i32 0, i32 0
// CHECK: [[T0:%.*]] = extractvalue { i64, i64 } [[CALL]], 0
@@ -234,7 +232,7 @@ typedef union {
TEST(union_het_fp)
// CHECK-LABEL: define dso_local swiftcc i64 @return_union_het_fp()
// CHECK: [[RET:%.*]] = alloca [[UNION:%.*]], align 8
-// CHECK: call void @llvm.memcpy{{.*}}(ptr align {{[0-9]+}} [[RET]]
+// CHECK: call void @llvm.memset{{.*}}(ptr align {{[0-9]+}} [[RET]]
// CHECK: [[GEP:%.*]] = getelementptr inbounds nuw { i64 }, ptr [[RET]], i32 0, i32 0
// CHECK: [[R0:%.*]] = load i64, ptr [[GEP]], align 8
// CHECK: ret i64 [[R0]]
diff --git a/clang/test/CodeGenObjC/designated-initializers.m b/clang/test/CodeGenObjC/designated-initializers.m
index a67f82e1afbea8..ce58f6c367338e 100644
--- a/clang/test/CodeGenObjC/designated-initializers.m
+++ b/clang/test/CodeGenObjC/designated-initializers.m
@@ -4,4 +4,4 @@
char L[3];
int M;
} overwrite_string[] = { { { @encode(void**) }, 1 }, [0].L[1] = 'x'};
-// CHECK: [3 x i8] c"^xv", i32 1
+// CHECK: [3 x i8] c"^xv", i8 0, i32 1
>From 56cad9ac66d4e63b6274e243e00200cd1be11787 Mon Sep 17 00:00:00 2001
From: Yabin Cui <yabinc at google.com>
Date: Fri, 27 Sep 2024 14:01:50 -0700
Subject: [PATCH 2/3] Zero init padding between bitfields in CGExprAgg.cpp
This isn't needed for CGExprConstant.cpp. Because an initialization
in one bit can prevent generating undef field in the const value.
---
clang/lib/CodeGen/CGExprAgg.cpp | 83 ++++++++++++-------
.../linux-kernel-struct-union-initializer2.c | 12 ++-
2 files changed, 65 insertions(+), 30 deletions(-)
diff --git a/clang/lib/CodeGen/CGExprAgg.cpp b/clang/lib/CodeGen/CGExprAgg.cpp
index 2bf67cee5fa926..3950e8206fb869 100644
--- a/clang/lib/CodeGen/CGExprAgg.cpp
+++ b/clang/lib/CodeGen/CGExprAgg.cpp
@@ -65,6 +65,9 @@ class AggExprEmitter : public StmtVisitor<AggExprEmitter> {
void withReturnValueSlot(const Expr *E,
llvm::function_ref<RValue(ReturnValueSlot)> Fn);
+ void DoZeroInitPadding(const Address BaseAddr, uint64_t StartBitOffset,
+ uint64_t EndBitOffset);
+
public:
AggExprEmitter(CodeGenFunction &cgf, AggValueSlot Dest, bool IsResultUnused)
: CGF(cgf), Builder(CGF.Builder), Dest(Dest),
@@ -1701,14 +1704,7 @@ void AggExprEmitter::VisitCXXParenListOrInitListExpr(
const bool ZeroInitPadding =
CGF.CGM.shouldZeroInitPadding() && !Dest.isZeroed();
- const Address BaseLoc = Dest.getAddress().withElementType(CGF.Int8Ty);
- auto DoZeroInitPadding = [&](CharUnits Offset, CharUnits Size) {
- if (Size.isPositive()) {
- Address Loc = CGF.Builder.CreateConstGEP(BaseLoc, Offset.getQuantity());
- llvm::Constant *SizeVal = CGF.Builder.getInt64(Size.getQuantity());
- CGF.Builder.CreateMemSet(Loc, CGF.Builder.getInt8(0), SizeVal, false);
- }
- };
+ const Address BaseAddr = Dest.getAddress().withElementType(CGF.Int8Ty);
if (record->isUnion()) {
// Only initialize one field of a union. The field itself is
@@ -1735,11 +1731,10 @@ void AggExprEmitter::VisitCXXParenListOrInitListExpr(
// Store the initializer into the field
EmitInitializationToLValue(InitExprs[0], FieldLoc);
if (ZeroInitPadding) {
- CharUnits TotalSize =
- Dest.getPreferredSize(CGF.getContext(), DestLV.getType());
- CharUnits FieldSize =
- CGF.getContext().getTypeSizeInChars(FieldLoc.getType());
- DoZeroInitPadding(FieldSize, TotalSize - FieldSize);
+ uint64_t TotalSize = CGF.getContext().toBits(
+ Dest.getPreferredSize(CGF.getContext(), DestLV.getType()));
+ uint64_t FieldSize = CGF.getContext().getTypeSize(FieldLoc.getType());
+ DoZeroInitPadding(BaseAddr, FieldSize, TotalSize);
}
} else {
// Default-initialize to null.
@@ -1754,7 +1749,7 @@ void AggExprEmitter::VisitCXXParenListOrInitListExpr(
// Here we iterate over the fields; this makes it simpler to both
// default-initialize fields and skip over unnamed fields.
const ASTRecordLayout &Layout = CGF.getContext().getASTRecordLayout(record);
- CharUnits SizeSoFar = CharUnits::Zero();
+ uint64_t LastFieldBitOffset = 0;
for (const auto *field : record->fields()) {
// We're done once we hit the flexible array member.
if (field->getType()->isIncompleteArrayType())
@@ -1773,22 +1768,15 @@ void AggExprEmitter::VisitCXXParenListOrInitListExpr(
if (ZeroInitPadding) {
uint64_t StartBitOffset = Layout.getFieldOffset(field->getFieldIndex());
- CharUnits StartOffset =
- CGF.getContext().toCharUnitsFromBits(StartBitOffset);
- DoZeroInitPadding(SizeSoFar, StartOffset - SizeSoFar);
+ DoZeroInitPadding(BaseAddr, LastFieldBitOffset, StartBitOffset);
if (!field->isBitField()) {
- CharUnits FieldSize =
- CGF.getContext().getTypeSizeInChars(field->getType());
- SizeSoFar = StartOffset + FieldSize;
+ LastFieldBitOffset =
+ StartBitOffset + CGF.getContext().getTypeSize(field->getType());
} else {
const CGRecordLayout &RL =
CGF.getTypes().getCGRecordLayout(field->getParent());
const CGBitFieldInfo &Info = RL.getBitFieldInfo(field);
- uint64_t EndBitOffset = StartBitOffset + Info.Size;
- SizeSoFar = CGF.getContext().toCharUnitsFromBits(EndBitOffset);
- if (EndBitOffset % CGF.getContext().getCharWidth() != 0) {
- SizeSoFar++;
- }
+ LastFieldBitOffset = StartBitOffset + Info.Size;
}
}
@@ -1818,9 +1806,48 @@ void AggExprEmitter::VisitCXXParenListOrInitListExpr(
}
}
if (ZeroInitPadding) {
- CharUnits TotalSize =
- Dest.getPreferredSize(CGF.getContext(), DestLV.getType());
- DoZeroInitPadding(SizeSoFar, TotalSize - SizeSoFar);
+ uint64_t TotalSize = CGF.getContext().toBits(
+ Dest.getPreferredSize(CGF.getContext(), DestLV.getType()));
+ DoZeroInitPadding(BaseAddr, LastFieldBitOffset, TotalSize);
+ }
+}
+
+void AggExprEmitter::DoZeroInitPadding(const Address BaseAddr,
+ uint64_t StartBitOffset,
+ uint64_t EndBitOffset) {
+ if (StartBitOffset >= EndBitOffset)
+ return;
+
+ auto InitBytes = [&](uint64_t Start, uint64_t End) {
+ Address Addr = CGF.Builder.CreateConstGEP(BaseAddr, Start);
+ llvm::Constant *SizeVal = CGF.Builder.getInt64(End - Start);
+ CGF.Builder.CreateMemSet(Addr, CGF.Builder.getInt8(0), SizeVal, false);
+ };
+ auto InitBits = [&](uint64_t Byte, uint64_t Start, uint64_t End) {
+ Address Addr = CGF.Builder.CreateConstGEP(BaseAddr, Byte);
+ llvm::Value *Val = Builder.CreateLoad(Addr);
+ Val = Builder.CreateAnd(Val, ~llvm::APInt::getBitsSet(8, Start, End));
+ Builder.CreateStore(Val, Addr);
+ };
+
+ uint64_t StartBit = StartBitOffset % 8;
+ uint64_t StartByte = StartBitOffset / 8;
+ uint64_t EndBit = EndBitOffset % 8;
+ uint64_t EndByte = EndBitOffset / 8;
+
+ if (StartByte < EndByte) {
+ if (StartBit != 0) {
+ InitBits(StartByte, StartBit, 8);
+ StartBit = 0;
+ StartByte++;
+ }
+ if (StartByte < EndByte) {
+ InitBytes(StartByte, EndByte);
+ StartByte = EndByte;
+ }
+ }
+ if (EndBit != 0) {
+ InitBits(EndByte, StartBit, EndBit);
}
}
diff --git a/clang/test/CodeGen/linux-kernel-struct-union-initializer2.c b/clang/test/CodeGen/linux-kernel-struct-union-initializer2.c
index bf8154601b4b5e..1f878ed48956eb 100644
--- a/clang/test/CodeGen/linux-kernel-struct-union-initializer2.c
+++ b/clang/test/CodeGen/linux-kernel-struct-union-initializer2.c
@@ -86,6 +86,10 @@ void test2(long long y) {
// CHECK-NEXT: [[BF_CLEAR:%.*]] = and i16 [[BF_LOAD]], -8
// CHECK-NEXT: [[BF_SET:%.*]] = or i16 [[BF_CLEAR]], [[BF_VALUE]]
// CHECK-NEXT: store i16 [[BF_SET]], ptr [[S]], align 4
+// CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[S]], i64 0
+// CHECK-NEXT: [[TMP3:%.*]] = load i8, ptr [[TMP2]], align 4
+// CHECK-NEXT: [[TMP4:%.*]] = and i8 [[TMP3]], 7
+// CHECK-NEXT: store i8 [[TMP4]], ptr [[TMP2]], align 4
// CHECK-NEXT: [[BF_LOAD1:%.*]] = load i16, ptr [[S]], align 4
// CHECK-NEXT: [[BF_CLEAR2:%.*]] = and i16 [[BF_LOAD1]], -16129
// CHECK-NEXT: [[BF_SET3:%.*]] = or i16 [[BF_CLEAR2]], 0
@@ -94,8 +98,8 @@ void test2(long long y) {
// CHECK-NEXT: [[BF_CLEAR5:%.*]] = and i16 [[BF_LOAD4]], 16383
// CHECK-NEXT: [[BF_SET6:%.*]] = or i16 [[BF_CLEAR5]], 0
// CHECK-NEXT: store i16 [[BF_SET6]], ptr [[S]], align 4
-// CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[S]], i64 2
-// CHECK-NEXT: call void @llvm.memset.p0.i64(ptr align 2 [[TMP2]], i8 0, i64 2, i1 false)
+// CHECK-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[S]], i64 2
+// CHECK-NEXT: call void @llvm.memset.p0.i64(ptr align 2 [[TMP5]], i8 0, i64 2, i1 false)
// CHECK-NEXT: [[I:%.*]] = getelementptr inbounds nuw [[STRUCT_S2]], ptr [[S]], i32 0, i32 1
// CHECK-NEXT: store i32 0, ptr [[I]], align 4
// CHECK-NEXT: ret void
@@ -164,6 +168,10 @@ void test5(int a, int b) {
// CHECK-NEXT: [[BF_CLEAR3:%.*]] = and i16 [[BF_LOAD2]], -2033
// CHECK-NEXT: [[BF_SET4:%.*]] = or i16 [[BF_CLEAR3]], 0
// CHECK-NEXT: store i16 [[BF_SET4]], ptr [[Z]], align 1
+// CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[S]], i64 2
+// CHECK-NEXT: [[TMP2:%.*]] = load i8, ptr [[TMP1]], align 1
+// CHECK-NEXT: [[TMP3:%.*]] = and i8 [[TMP2]], 7
+// CHECK-NEXT: store i8 [[TMP3]], ptr [[TMP1]], align 1
// CHECK-NEXT: ret void
//
void test6(char x) {
>From 5ffefd5133fcf76f99f01553544ae27f18ab271e Mon Sep 17 00:00:00 2001
From: Yabin Cui <yabinc at google.com>
Date: Fri, 4 Oct 2024 14:23:09 -0700
Subject: [PATCH 3/3] Init paddings between bitfields by storing 0 to
StorageSize
When seeing a bitfield, store 0 to its [StorageOffset,
StorageOffset+StorageSize]. The store width is StorageSize,
making it easier to be merged with other bitfield stores.
And we don't need to worry about big/little endian problem.
---
clang/lib/CodeGen/CGExprAgg.cpp | 98 +++++++++----------
.../linux-kernel-struct-union-initializer2.c | 15 +--
2 files changed, 51 insertions(+), 62 deletions(-)
diff --git a/clang/lib/CodeGen/CGExprAgg.cpp b/clang/lib/CodeGen/CGExprAgg.cpp
index 3950e8206fb869..8f2e4c3712799f 100644
--- a/clang/lib/CodeGen/CGExprAgg.cpp
+++ b/clang/lib/CodeGen/CGExprAgg.cpp
@@ -65,8 +65,8 @@ class AggExprEmitter : public StmtVisitor<AggExprEmitter> {
void withReturnValueSlot(const Expr *E,
llvm::function_ref<RValue(ReturnValueSlot)> Fn);
- void DoZeroInitPadding(const Address BaseAddr, uint64_t StartBitOffset,
- uint64_t EndBitOffset);
+ void DoZeroInitPadding(uint64_t &PaddingStart, uint64_t PaddingEnd,
+ const FieldDecl *NextField);
public:
AggExprEmitter(CodeGenFunction &cgf, AggValueSlot Dest, bool IsResultUnused)
@@ -1704,7 +1704,6 @@ void AggExprEmitter::VisitCXXParenListOrInitListExpr(
const bool ZeroInitPadding =
CGF.CGM.shouldZeroInitPadding() && !Dest.isZeroed();
- const Address BaseAddr = Dest.getAddress().withElementType(CGF.Int8Ty);
if (record->isUnion()) {
// Only initialize one field of a union. The field itself is
@@ -1734,7 +1733,7 @@ void AggExprEmitter::VisitCXXParenListOrInitListExpr(
uint64_t TotalSize = CGF.getContext().toBits(
Dest.getPreferredSize(CGF.getContext(), DestLV.getType()));
uint64_t FieldSize = CGF.getContext().getTypeSize(FieldLoc.getType());
- DoZeroInitPadding(BaseAddr, FieldSize, TotalSize);
+ DoZeroInitPadding(FieldSize, TotalSize, nullptr);
}
} else {
// Default-initialize to null.
@@ -1749,7 +1748,8 @@ void AggExprEmitter::VisitCXXParenListOrInitListExpr(
// Here we iterate over the fields; this makes it simpler to both
// default-initialize fields and skip over unnamed fields.
const ASTRecordLayout &Layout = CGF.getContext().getASTRecordLayout(record);
- uint64_t LastFieldBitOffset = 0;
+ uint64_t PaddingStart = 0;
+
for (const auto *field : record->fields()) {
// We're done once we hit the flexible array member.
if (field->getType()->isIncompleteArrayType())
@@ -1766,19 +1766,9 @@ void AggExprEmitter::VisitCXXParenListOrInitListExpr(
CGF.getTypes().isZeroInitializable(ExprToVisit->getType()))
break;
- if (ZeroInitPadding) {
- uint64_t StartBitOffset = Layout.getFieldOffset(field->getFieldIndex());
- DoZeroInitPadding(BaseAddr, LastFieldBitOffset, StartBitOffset);
- if (!field->isBitField()) {
- LastFieldBitOffset =
- StartBitOffset + CGF.getContext().getTypeSize(field->getType());
- } else {
- const CGRecordLayout &RL =
- CGF.getTypes().getCGRecordLayout(field->getParent());
- const CGBitFieldInfo &Info = RL.getBitFieldInfo(field);
- LastFieldBitOffset = StartBitOffset + Info.Size;
- }
- }
+ if (ZeroInitPadding)
+ DoZeroInitPadding(PaddingStart,
+ Layout.getFieldOffset(field->getFieldIndex()), field);
LValue LV = CGF.EmitLValueForFieldInitialization(DestLV, field);
// We never generate write-barries for initialized fields.
@@ -1808,47 +1798,51 @@ void AggExprEmitter::VisitCXXParenListOrInitListExpr(
if (ZeroInitPadding) {
uint64_t TotalSize = CGF.getContext().toBits(
Dest.getPreferredSize(CGF.getContext(), DestLV.getType()));
- DoZeroInitPadding(BaseAddr, LastFieldBitOffset, TotalSize);
+ DoZeroInitPadding(PaddingStart, TotalSize, nullptr);
}
}
-void AggExprEmitter::DoZeroInitPadding(const Address BaseAddr,
- uint64_t StartBitOffset,
- uint64_t EndBitOffset) {
- if (StartBitOffset >= EndBitOffset)
- return;
-
- auto InitBytes = [&](uint64_t Start, uint64_t End) {
- Address Addr = CGF.Builder.CreateConstGEP(BaseAddr, Start);
- llvm::Constant *SizeVal = CGF.Builder.getInt64(End - Start);
- CGF.Builder.CreateMemSet(Addr, CGF.Builder.getInt8(0), SizeVal, false);
- };
- auto InitBits = [&](uint64_t Byte, uint64_t Start, uint64_t End) {
- Address Addr = CGF.Builder.CreateConstGEP(BaseAddr, Byte);
- llvm::Value *Val = Builder.CreateLoad(Addr);
- Val = Builder.CreateAnd(Val, ~llvm::APInt::getBitsSet(8, Start, End));
- Builder.CreateStore(Val, Addr);
+void AggExprEmitter::DoZeroInitPadding(uint64_t &PaddingStart,
+ uint64_t PaddingEnd,
+ const FieldDecl *NextField) {
+
+ auto InitBytes = [&](uint64_t StartBit, uint64_t EndBit) {
+ CharUnits Start = CGF.getContext().toCharUnitsFromBits(StartBit);
+ CharUnits End = CGF.getContext().toCharUnitsFromBits(EndBit);
+ Address Addr = Dest.getAddress().withElementType(CGF.CharTy);
+ if (!Start.isZero())
+ Addr = Builder.CreateConstGEP(Addr, Start.getQuantity());
+ llvm::Constant *SizeVal = Builder.getInt64((End - Start).getQuantity());
+ CGF.Builder.CreateMemSet(Addr, Builder.getInt8(0), SizeVal, false);
};
- uint64_t StartBit = StartBitOffset % 8;
- uint64_t StartByte = StartBitOffset / 8;
- uint64_t EndBit = EndBitOffset % 8;
- uint64_t EndByte = EndBitOffset / 8;
-
- if (StartByte < EndByte) {
- if (StartBit != 0) {
- InitBits(StartByte, StartBit, 8);
- StartBit = 0;
- StartByte++;
- }
- if (StartByte < EndByte) {
- InitBytes(StartByte, EndByte);
- StartByte = EndByte;
+ if (NextField != nullptr && NextField->isBitField()) {
+ // For bitfield, zero init StorageSize before storing the bits. So we don't
+ // need to handle big/little endian.
+ const CGRecordLayout &RL =
+ CGF.getTypes().getCGRecordLayout(NextField->getParent());
+ const CGBitFieldInfo &Info = RL.getBitFieldInfo(NextField);
+ uint64_t StorageStart = CGF.getContext().toBits(Info.StorageOffset);
+ if (StorageStart + Info.StorageSize > PaddingStart) {
+ if (StorageStart > PaddingStart)
+ InitBytes(PaddingStart, StorageStart);
+ Address Addr = Dest.getAddress();
+ if (!Info.StorageOffset.isZero())
+ Addr = Builder.CreateConstGEP(Addr.withElementType(CGF.CharTy),
+ Info.StorageOffset.getQuantity());
+ Addr = Addr.withElementType(
+ llvm::Type::getIntNTy(CGF.getLLVMContext(), Info.StorageSize));
+ Builder.CreateStore(Builder.getIntN(Info.StorageSize, 0), Addr);
+ PaddingStart = StorageStart + Info.StorageSize;
}
+ return;
}
- if (EndBit != 0) {
- InitBits(EndByte, StartBit, EndBit);
- }
+
+ if (PaddingStart < PaddingEnd)
+ InitBytes(PaddingStart, PaddingEnd);
+ if (NextField != nullptr)
+ PaddingStart =
+ PaddingEnd + CGF.getContext().getTypeSize(NextField->getType());
}
void AggExprEmitter::VisitArrayInitLoopExpr(const ArrayInitLoopExpr *E,
diff --git a/clang/test/CodeGen/linux-kernel-struct-union-initializer2.c b/clang/test/CodeGen/linux-kernel-struct-union-initializer2.c
index 1f878ed48956eb..df53d194c6aad7 100644
--- a/clang/test/CodeGen/linux-kernel-struct-union-initializer2.c
+++ b/clang/test/CodeGen/linux-kernel-struct-union-initializer2.c
@@ -79,6 +79,7 @@ void test2(long long y) {
// CHECK-NEXT: [[B_ADDR:%.*]] = alloca i8, align 1
// CHECK-NEXT: [[S:%.*]] = alloca [[STRUCT_S2:%.*]], align 4
// CHECK-NEXT: store i8 [[B]], ptr [[B_ADDR]], align 1
+// CHECK-NEXT: store i16 0, ptr [[S]], align 4
// CHECK-NEXT: [[TMP0:%.*]] = load i8, ptr [[B_ADDR]], align 1
// CHECK-NEXT: [[TMP1:%.*]] = zext i8 [[TMP0]] to i16
// CHECK-NEXT: [[BF_LOAD:%.*]] = load i16, ptr [[S]], align 4
@@ -86,10 +87,6 @@ void test2(long long y) {
// CHECK-NEXT: [[BF_CLEAR:%.*]] = and i16 [[BF_LOAD]], -8
// CHECK-NEXT: [[BF_SET:%.*]] = or i16 [[BF_CLEAR]], [[BF_VALUE]]
// CHECK-NEXT: store i16 [[BF_SET]], ptr [[S]], align 4
-// CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[S]], i64 0
-// CHECK-NEXT: [[TMP3:%.*]] = load i8, ptr [[TMP2]], align 4
-// CHECK-NEXT: [[TMP4:%.*]] = and i8 [[TMP3]], 7
-// CHECK-NEXT: store i8 [[TMP4]], ptr [[TMP2]], align 4
// CHECK-NEXT: [[BF_LOAD1:%.*]] = load i16, ptr [[S]], align 4
// CHECK-NEXT: [[BF_CLEAR2:%.*]] = and i16 [[BF_LOAD1]], -16129
// CHECK-NEXT: [[BF_SET3:%.*]] = or i16 [[BF_CLEAR2]], 0
@@ -98,8 +95,8 @@ void test2(long long y) {
// CHECK-NEXT: [[BF_CLEAR5:%.*]] = and i16 [[BF_LOAD4]], 16383
// CHECK-NEXT: [[BF_SET6:%.*]] = or i16 [[BF_CLEAR5]], 0
// CHECK-NEXT: store i16 [[BF_SET6]], ptr [[S]], align 4
-// CHECK-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[S]], i64 2
-// CHECK-NEXT: call void @llvm.memset.p0.i64(ptr align 2 [[TMP5]], i8 0, i64 2, i1 false)
+// CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[S]], i64 2
+// CHECK-NEXT: call void @llvm.memset.p0.i64(ptr align 2 [[TMP2]], i8 0, i64 2, i1 false)
// CHECK-NEXT: [[I:%.*]] = getelementptr inbounds nuw [[STRUCT_S2]], ptr [[S]], i32 0, i32 1
// CHECK-NEXT: store i32 0, ptr [[I]], align 4
// CHECK-NEXT: ret void
@@ -158,6 +155,8 @@ void test5(int a, int b) {
// CHECK-NEXT: [[X1:%.*]] = getelementptr inbounds nuw [[STRUCT_S5]], ptr [[S]], i32 0, i32 0
// CHECK-NEXT: [[TMP0:%.*]] = load i8, ptr [[X_ADDR]], align 1
// CHECK-NEXT: store i8 [[TMP0]], ptr [[X1]], align 1
+// CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[S]], i64 1
+// CHECK-NEXT: store i16 0, ptr [[TMP1]], align 1
// CHECK-NEXT: [[Y:%.*]] = getelementptr inbounds nuw [[STRUCT_S5]], ptr [[S]], i32 0, i32 1
// CHECK-NEXT: [[BF_LOAD:%.*]] = load i16, ptr [[Y]], align 1
// CHECK-NEXT: [[BF_CLEAR:%.*]] = and i16 [[BF_LOAD]], -16
@@ -168,10 +167,6 @@ void test5(int a, int b) {
// CHECK-NEXT: [[BF_CLEAR3:%.*]] = and i16 [[BF_LOAD2]], -2033
// CHECK-NEXT: [[BF_SET4:%.*]] = or i16 [[BF_CLEAR3]], 0
// CHECK-NEXT: store i16 [[BF_SET4]], ptr [[Z]], align 1
-// CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[S]], i64 2
-// CHECK-NEXT: [[TMP2:%.*]] = load i8, ptr [[TMP1]], align 1
-// CHECK-NEXT: [[TMP3:%.*]] = and i8 [[TMP2]], 7
-// CHECK-NEXT: store i8 [[TMP3]], ptr [[TMP1]], align 1
// CHECK-NEXT: ret void
//
void test6(char x) {
More information about the cfe-commits
mailing list