CodeGenTBAA.cpp revision 8a06d8fdefce6b4ed4438bb80f60cdd2e979a94f 
1//===--- CodeGenTypes.cpp - TBAA information for LLVM CodeGen -------------===//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// This is the code that manages TBAA information and defines the TBAA policy
11// for the optimizer to use. Relevant standards text includes:
12//
13//   C99 6.5p7
14//   C++ [basic.lval] (p10 in n3126, p15 in some earlier versions)
15//
16//===----------------------------------------------------------------------===//
17
18#include "CodeGenTBAA.h"
19#include "clang/AST/ASTContext.h"
20#include "clang/AST/Attr.h"
21#include "clang/AST/Mangle.h"
22#include "clang/AST/RecordLayout.h"
23#include "clang/Frontend/CodeGenOptions.h"
24#include "llvm/ADT/SmallSet.h"
25#include "llvm/IR/Constants.h"
26#include "llvm/IR/LLVMContext.h"
27#include "llvm/IR/Metadata.h"
28#include "llvm/IR/Type.h"
29using namespace clang;
30using namespace CodeGen;
31
32CodeGenTBAA::CodeGenTBAA(ASTContext &Ctx, llvm::LLVMContext& VMContext,
33                         const CodeGenOptions &CGO,
34                         const LangOptions &Features, MangleContext &MContext)
35  : Context(Ctx), CodeGenOpts(CGO), Features(Features), MContext(MContext),
36    MDHelper(VMContext), Root(0), Char(0) {
37}
38
39CodeGenTBAA::~CodeGenTBAA() {
40}
41
42llvm::MDNode *CodeGenTBAA::getRoot() {
43  // Define the root of the tree. This identifies the tree, so that
44  // if our LLVM IR is linked with LLVM IR from a different front-end
45  // (or a different version of this front-end), their TBAA trees will
46  // remain distinct, and the optimizer will treat them conservatively.
47  if (!Root)
48    Root = MDHelper.createTBAARoot("Simple C/C++ TBAA");
49
50  return Root;
51}
52
53// For both scalar TBAA and struct-path aware TBAA, the scalar type has the
54// same format: name, parent node, and offset.
55llvm::MDNode *CodeGenTBAA::createTBAAScalarType(StringRef Name,
56                                                llvm::MDNode *Parent) {
57  return MDHelper.createTBAAScalarTypeNode(Name, Parent);
58}
59
60llvm::MDNode *CodeGenTBAA::getChar() {
61  // Define the root of the tree for user-accessible memory. C and C++
62  // give special powers to char and certain similar types. However,
63  // these special powers only cover user-accessible memory, and doesn't
64  // include things like vtables.
65  if (!Char)
66    Char = createTBAAScalarType("omnipotent char", getRoot());
67
68  return Char;
69}
70
71static bool TypeHasMayAlias(QualType QTy) {
72  // Tagged types have declarations, and therefore may have attributes.
73  if (const TagType *TTy = dyn_cast<TagType>(QTy))
74    return TTy->getDecl()->hasAttr<MayAliasAttr>();
75
76  // Typedef types have declarations, and therefore may have attributes.
77  if (const TypedefType *TTy = dyn_cast<TypedefType>(QTy)) {
78    if (TTy->getDecl()->hasAttr<MayAliasAttr>())
79      return true;
80    // Also, their underlying types may have relevant attributes.
81    return TypeHasMayAlias(TTy->desugar());
82  }
83
84  return false;
85}
86
87llvm::MDNode *
88CodeGenTBAA::getTBAAInfo(QualType QTy) {
89  // At -O0 TBAA is not emitted for regular types.
90  if (CodeGenOpts.OptimizationLevel == 0 || CodeGenOpts.RelaxedAliasing)
91    return NULL;
92
93  // If the type has the may_alias attribute (even on a typedef), it is
94  // effectively in the general char alias class.
95  if (TypeHasMayAlias(QTy))
96    return getChar();
97
98  const Type *Ty = Context.getCanonicalType(QTy).getTypePtr();
99
100  if (llvm::MDNode *N = MetadataCache[Ty])
101    return N;
102
103  // Handle builtin types.
104  if (const BuiltinType *BTy = dyn_cast<BuiltinType>(Ty)) {
105    switch (BTy->getKind()) {
106    // Character types are special and can alias anything.
107    // In C++, this technically only includes "char" and "unsigned char",
108    // and not "signed char". In C, it includes all three. For now,
109    // the risk of exploiting this detail in C++ seems likely to outweigh
110    // the benefit.
111    case BuiltinType::Char_U:
112    case BuiltinType::Char_S:
113    case BuiltinType::UChar:
114    case BuiltinType::SChar:
115      return getChar();
116
117    // Unsigned types can alias their corresponding signed types.
118    case BuiltinType::UShort:
119      return getTBAAInfo(Context.ShortTy);
120    case BuiltinType::UInt:
121      return getTBAAInfo(Context.IntTy);
122    case BuiltinType::ULong:
123      return getTBAAInfo(Context.LongTy);
124    case BuiltinType::ULongLong:
125      return getTBAAInfo(Context.LongLongTy);
126    case BuiltinType::UInt128:
127      return getTBAAInfo(Context.Int128Ty);
128
129    // Treat all other builtin types as distinct types. This includes
130    // treating wchar_t, char16_t, and char32_t as distinct from their
131    // "underlying types".
132    default:
133      return MetadataCache[Ty] =
134        createTBAAScalarType(BTy->getName(Features), getChar());
135    }
136  }
137
138  // Handle pointers.
139  // TODO: Implement C++'s type "similarity" and consider dis-"similar"
140  // pointers distinct.
141  if (Ty->isPointerType())
142    return MetadataCache[Ty] = createTBAAScalarType("any pointer",
143                                                    getChar());
144
145  // Enum types are distinct types. In C++ they have "underlying types",
146  // however they aren't related for TBAA.
147  if (const EnumType *ETy = dyn_cast<EnumType>(Ty)) {
148    // In C++ mode, types have linkage, so we can rely on the ODR and
149    // on their mangled names, if they're external.
150    // TODO: Is there a way to get a program-wide unique name for a
151    // decl with local linkage or no linkage?
152    if (!Features.CPlusPlus || !ETy->getDecl()->isExternallyVisible())
153      return MetadataCache[Ty] = getChar();
154
155    // TODO: This is using the RTTI name. Is there a better way to get
156    // a unique string for a type?
157    SmallString<256> OutName;
158    llvm::raw_svector_ostream Out(OutName);
159    MContext.mangleCXXRTTIName(QualType(ETy, 0), Out);
160    Out.flush();
161    return MetadataCache[Ty] = createTBAAScalarType(OutName, getChar());
162  }
163
164  // For now, handle any other kind of type conservatively.
165  return MetadataCache[Ty] = getChar();
166}
167
168llvm::MDNode *CodeGenTBAA::getTBAAInfoForVTablePtr() {
169  return createTBAAScalarType("vtable pointer", getRoot());
170}
171
172bool
173CodeGenTBAA::CollectFields(uint64_t BaseOffset,
174                           QualType QTy,
175                           SmallVectorImpl<llvm::MDBuilder::TBAAStructField> &
176                             Fields,
177                           bool MayAlias) {
178  /* Things not handled yet include: C++ base classes, bitfields, */
179
180  if (const RecordType *TTy = QTy->getAs<RecordType>()) {
181    const RecordDecl *RD = TTy->getDecl()->getDefinition();
182    if (RD->hasFlexibleArrayMember())
183      return false;
184
185    // TODO: Handle C++ base classes.
186    if (const CXXRecordDecl *Decl = dyn_cast<CXXRecordDecl>(RD))
187      if (Decl->bases_begin() != Decl->bases_end())
188        return false;
189
190    const ASTRecordLayout &Layout = Context.getASTRecordLayout(RD);
191
192    unsigned idx = 0;
193    for (RecordDecl::field_iterator i = RD->field_begin(),
194         e = RD->field_end(); i != e; ++i, ++idx) {
195      uint64_t Offset = BaseOffset +
196                        Layout.getFieldOffset(idx) / Context.getCharWidth();
197      QualType FieldQTy = i->getType();
198      if (!CollectFields(Offset, FieldQTy, Fields,
199                         MayAlias || TypeHasMayAlias(FieldQTy)))
200        return false;
201    }
202    return true;
203  }
204
205  /* Otherwise, treat whatever it is as a field. */
206  uint64_t Offset = BaseOffset;
207  uint64_t Size = Context.getTypeSizeInChars(QTy).getQuantity();
208  llvm::MDNode *TBAAInfo = MayAlias ? getChar() : getTBAAInfo(QTy);
209  llvm::MDNode *TBAATag = getTBAAScalarTagInfo(TBAAInfo);
210  Fields.push_back(llvm::MDBuilder::TBAAStructField(Offset, Size, TBAATag));
211  return true;
212}
213
214llvm::MDNode *
215CodeGenTBAA::getTBAAStructInfo(QualType QTy) {
216  const Type *Ty = Context.getCanonicalType(QTy).getTypePtr();
217
218  if (llvm::MDNode *N = StructMetadataCache[Ty])
219    return N;
220
221  SmallVector<llvm::MDBuilder::TBAAStructField, 4> Fields;
222  if (CollectFields(0, QTy, Fields, TypeHasMayAlias(QTy)))
223    return MDHelper.createTBAAStructNode(Fields);
224
225  // For now, handle any other kind of type conservatively.
226  return StructMetadataCache[Ty] = NULL;
227}
228
229/// Check if the given type can be handled by path-aware TBAA.
230static bool isTBAAPathStruct(QualType QTy) {
231  if (const RecordType *TTy = QTy->getAs<RecordType>()) {
232    const RecordDecl *RD = TTy->getDecl()->getDefinition();
233    if (RD->hasFlexibleArrayMember())
234      return false;
235    // RD can be struct, union, class, interface or enum.
236    // For now, we only handle struct and class.
237    if (RD->isStruct() || RD->isClass())
238      return true;
239  }
240  return false;
241}
242
243llvm::MDNode *
244CodeGenTBAA::getTBAAStructTypeInfo(QualType QTy) {
245  const Type *Ty = Context.getCanonicalType(QTy).getTypePtr();
246  assert(isTBAAPathStruct(QTy));
247
248  if (llvm::MDNode *N = StructTypeMetadataCache[Ty])
249    return N;
250
251  if (const RecordType *TTy = QTy->getAs<RecordType>()) {
252    const RecordDecl *RD = TTy->getDecl()->getDefinition();
253
254    const ASTRecordLayout &Layout = Context.getASTRecordLayout(RD);
255    SmallVector <std::pair<llvm::MDNode*, uint64_t>, 4> Fields;
256    unsigned idx = 0;
257    for (RecordDecl::field_iterator i = RD->field_begin(),
258         e = RD->field_end(); i != e; ++i, ++idx) {
259      QualType FieldQTy = i->getType();
260      llvm::MDNode *FieldNode;
261      if (isTBAAPathStruct(FieldQTy))
262        FieldNode = getTBAAStructTypeInfo(FieldQTy);
263      else
264        FieldNode = getTBAAInfo(FieldQTy);
265      if (!FieldNode)
266        return StructTypeMetadataCache[Ty] = NULL;
267      Fields.push_back(std::make_pair(
268          FieldNode, Layout.getFieldOffset(idx) / Context.getCharWidth()));
269    }
270
271    // TODO: This is using the RTTI name. Is there a better way to get
272    // a unique string for a type?
273    SmallString<256> OutName;
274    if (Features.CPlusPlus) {
275      // Don't use mangleCXXRTTIName for C code.
276      llvm::raw_svector_ostream Out(OutName);
277      MContext.mangleCXXRTTIName(QualType(Ty, 0), Out);
278      Out.flush();
279    } else {
280      OutName = RD->getName();
281    }
282    // Create the struct type node with a vector of pairs (offset, type).
283    return StructTypeMetadataCache[Ty] =
284      MDHelper.createTBAAStructTypeNode(OutName, Fields);
285  }
286
287  return StructMetadataCache[Ty] = NULL;
288}
289
290/// Return a TBAA tag node for both scalar TBAA and struct-path aware TBAA.
291llvm::MDNode *
292CodeGenTBAA::getTBAAStructTagInfo(QualType BaseQTy, llvm::MDNode *AccessNode,
293                                  uint64_t Offset) {
294  if (!CodeGenOpts.StructPathTBAA)
295    return getTBAAScalarTagInfo(AccessNode);
296
297  const Type *BTy = Context.getCanonicalType(BaseQTy).getTypePtr();
298  TBAAPathTag PathTag = TBAAPathTag(BTy, AccessNode, Offset);
299  if (llvm::MDNode *N = StructTagMetadataCache[PathTag])
300    return N;
301
302  llvm::MDNode *BNode = 0;
303  if (isTBAAPathStruct(BaseQTy))
304    BNode  = getTBAAStructTypeInfo(BaseQTy);
305  if (!BNode)
306    return StructTagMetadataCache[PathTag] =
307       MDHelper.createTBAAStructTagNode(AccessNode, AccessNode, 0);
308
309  return StructTagMetadataCache[PathTag] =
310    MDHelper.createTBAAStructTagNode(BNode, AccessNode, Offset);
311}
312
313llvm::MDNode *
314CodeGenTBAA::getTBAAScalarTagInfo(llvm::MDNode *AccessNode) {
315  if (llvm::MDNode *N = ScalarTagMetadataCache[AccessNode])
316    return N;
317
318  return ScalarTagMetadataCache[AccessNode] =
319    MDHelper.createTBAAStructTagNode(AccessNode, AccessNode, 0);
320}
321