xref: /freebsd-src/contrib/llvm-project/clang/lib/AST/Interp/ByteCodeEmitter.cpp (revision 7a6dacaca14b62ca4b74406814becb87a3fefac0)
1a7dea167SDimitry Andric //===--- ByteCodeEmitter.cpp - Instruction emitter for the VM ---*- C++ -*-===//
2a7dea167SDimitry Andric //
3a7dea167SDimitry Andric // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4a7dea167SDimitry Andric // See https://llvm.org/LICENSE.txt for license information.
5a7dea167SDimitry Andric // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6a7dea167SDimitry Andric //
7a7dea167SDimitry Andric //===----------------------------------------------------------------------===//
8a7dea167SDimitry Andric 
9a7dea167SDimitry Andric #include "ByteCodeEmitter.h"
105f757f3fSDimitry Andric #include "ByteCodeGenError.h"
11a7dea167SDimitry Andric #include "Context.h"
1206c3fb27SDimitry Andric #include "Floating.h"
13a7dea167SDimitry Andric #include "Opcode.h"
14a7dea167SDimitry Andric #include "Program.h"
1506c3fb27SDimitry Andric #include "clang/AST/ASTLambda.h"
16a7dea167SDimitry Andric #include "clang/AST/DeclCXX.h"
175f757f3fSDimitry Andric #include "clang/Basic/Builtins.h"
18349cc55cSDimitry Andric #include <type_traits>
19a7dea167SDimitry Andric 
20a7dea167SDimitry Andric using namespace clang;
21a7dea167SDimitry Andric using namespace clang::interp;
22a7dea167SDimitry Andric 
23*7a6dacacSDimitry Andric Function *ByteCodeEmitter::compileFunc(const FunctionDecl *FuncDecl) {
24a7dea167SDimitry Andric   // Set up argument indices.
25a7dea167SDimitry Andric   unsigned ParamOffset = 0;
26a7dea167SDimitry Andric   SmallVector<PrimType, 8> ParamTypes;
2706c3fb27SDimitry Andric   SmallVector<unsigned, 8> ParamOffsets;
28a7dea167SDimitry Andric   llvm::DenseMap<unsigned, Function::ParamDescriptor> ParamDescriptors;
29a7dea167SDimitry Andric 
30bdd1243dSDimitry Andric   // If the return is not a primitive, a pointer to the storage where the
31bdd1243dSDimitry Andric   // value is initialized in is passed as the first argument. See 'RVO'
32bdd1243dSDimitry Andric   // elsewhere in the code.
33bdd1243dSDimitry Andric   QualType Ty = FuncDecl->getReturnType();
34bdd1243dSDimitry Andric   bool HasRVO = false;
35a7dea167SDimitry Andric   if (!Ty->isVoidType() && !Ctx.classify(Ty)) {
36bdd1243dSDimitry Andric     HasRVO = true;
37bdd1243dSDimitry Andric     ParamTypes.push_back(PT_Ptr);
3806c3fb27SDimitry Andric     ParamOffsets.push_back(ParamOffset);
39bdd1243dSDimitry Andric     ParamOffset += align(primSize(PT_Ptr));
40bdd1243dSDimitry Andric   }
41bdd1243dSDimitry Andric 
42bdd1243dSDimitry Andric   // If the function decl is a member decl, the next parameter is
43bdd1243dSDimitry Andric   // the 'this' pointer. This parameter is pop()ed from the
44bdd1243dSDimitry Andric   // InterpStack when calling the function.
45bdd1243dSDimitry Andric   bool HasThisPointer = false;
4606c3fb27SDimitry Andric   if (const auto *MD = dyn_cast<CXXMethodDecl>(FuncDecl)) {
475f757f3fSDimitry Andric     if (MD->isImplicitObjectMemberFunction()) {
48bdd1243dSDimitry Andric       HasThisPointer = true;
49a7dea167SDimitry Andric       ParamTypes.push_back(PT_Ptr);
5006c3fb27SDimitry Andric       ParamOffsets.push_back(ParamOffset);
51a7dea167SDimitry Andric       ParamOffset += align(primSize(PT_Ptr));
52a7dea167SDimitry Andric     }
53a7dea167SDimitry Andric 
5406c3fb27SDimitry Andric     // Set up lambda capture to closure record field mapping.
5506c3fb27SDimitry Andric     if (isLambdaCallOperator(MD)) {
5606c3fb27SDimitry Andric       const Record *R = P.getOrCreateRecord(MD->getParent());
5706c3fb27SDimitry Andric       llvm::DenseMap<const ValueDecl *, FieldDecl *> LC;
5806c3fb27SDimitry Andric       FieldDecl *LTC;
5906c3fb27SDimitry Andric 
6006c3fb27SDimitry Andric       MD->getParent()->getCaptureFields(LC, LTC);
6106c3fb27SDimitry Andric 
6206c3fb27SDimitry Andric       for (auto Cap : LC) {
635f757f3fSDimitry Andric         // Static lambdas cannot have any captures. If this one does,
645f757f3fSDimitry Andric         // it has already been diagnosed and we can only ignore it.
655f757f3fSDimitry Andric         if (MD->isStatic())
665f757f3fSDimitry Andric           return nullptr;
675f757f3fSDimitry Andric 
6806c3fb27SDimitry Andric         unsigned Offset = R->getField(Cap.second)->Offset;
6906c3fb27SDimitry Andric         this->LambdaCaptures[Cap.first] = {
7006c3fb27SDimitry Andric             Offset, Cap.second->getType()->isReferenceType()};
7106c3fb27SDimitry Andric       }
725f757f3fSDimitry Andric       if (LTC)
735f757f3fSDimitry Andric         this->LambdaThisCapture = R->getField(LTC)->Offset;
7406c3fb27SDimitry Andric     }
7506c3fb27SDimitry Andric   }
7606c3fb27SDimitry Andric 
77a7dea167SDimitry Andric   // Assign descriptors to all parameters.
78a7dea167SDimitry Andric   // Composite objects are lowered to pointers.
79bdd1243dSDimitry Andric   for (const ParmVarDecl *PD : FuncDecl->parameters()) {
805f757f3fSDimitry Andric     std::optional<PrimType> T = Ctx.classify(PD->getType());
815f757f3fSDimitry Andric     PrimType PT = T.value_or(PT_Ptr);
825f757f3fSDimitry Andric     Descriptor *Desc = P.createDescriptor(PD, PT);
835f757f3fSDimitry Andric     ParamDescriptors.insert({ParamOffset, {PT, Desc}});
845f757f3fSDimitry Andric     Params.insert({PD, {ParamOffset, T != std::nullopt}});
8506c3fb27SDimitry Andric     ParamOffsets.push_back(ParamOffset);
865f757f3fSDimitry Andric     ParamOffset += align(primSize(PT));
875f757f3fSDimitry Andric     ParamTypes.push_back(PT);
88a7dea167SDimitry Andric   }
89a7dea167SDimitry Andric 
9006c3fb27SDimitry Andric   // Create a handle over the emitted code.
9106c3fb27SDimitry Andric   Function *Func = P.getFunction(FuncDecl);
925f757f3fSDimitry Andric   if (!Func) {
935f757f3fSDimitry Andric     bool IsUnevaluatedBuiltin = false;
945f757f3fSDimitry Andric     if (unsigned BI = FuncDecl->getBuiltinID())
955f757f3fSDimitry Andric       IsUnevaluatedBuiltin = Ctx.getASTContext().BuiltinInfo.isUnevaluated(BI);
965f757f3fSDimitry Andric 
975f757f3fSDimitry Andric     Func =
985f757f3fSDimitry Andric         P.createFunction(FuncDecl, ParamOffset, std::move(ParamTypes),
995f757f3fSDimitry Andric                          std::move(ParamDescriptors), std::move(ParamOffsets),
1005f757f3fSDimitry Andric                          HasThisPointer, HasRVO, IsUnevaluatedBuiltin);
1015f757f3fSDimitry Andric   }
102bdd1243dSDimitry Andric 
103bdd1243dSDimitry Andric   assert(Func);
10406c3fb27SDimitry Andric   // For not-yet-defined functions, we only create a Function instance and
10506c3fb27SDimitry Andric   // compile their body later.
1065f757f3fSDimitry Andric   if (!FuncDecl->isDefined()) {
1075f757f3fSDimitry Andric     Func->setDefined(false);
108bdd1243dSDimitry Andric     return Func;
1095f757f3fSDimitry Andric   }
1105f757f3fSDimitry Andric 
1115f757f3fSDimitry Andric   Func->setDefined(true);
1125f757f3fSDimitry Andric 
1135f757f3fSDimitry Andric   // Lambda static invokers are a special case that we emit custom code for.
1145f757f3fSDimitry Andric   bool IsEligibleForCompilation = false;
1155f757f3fSDimitry Andric   if (const auto *MD = dyn_cast<CXXMethodDecl>(FuncDecl))
1165f757f3fSDimitry Andric     IsEligibleForCompilation = MD->isLambdaStaticInvoker();
1175f757f3fSDimitry Andric   if (!IsEligibleForCompilation)
1185f757f3fSDimitry Andric     IsEligibleForCompilation = FuncDecl->isConstexpr();
119bdd1243dSDimitry Andric 
120a7dea167SDimitry Andric   // Compile the function body.
1215f757f3fSDimitry Andric   if (!IsEligibleForCompilation || !visitFunc(FuncDecl)) {
122bdd1243dSDimitry Andric     Func->setIsFullyCompiled(true);
123a7dea167SDimitry Andric     return Func;
124bdd1243dSDimitry Andric   }
1255f757f3fSDimitry Andric 
126a7dea167SDimitry Andric   // Create scopes from descriptors.
127a7dea167SDimitry Andric   llvm::SmallVector<Scope, 2> Scopes;
128a7dea167SDimitry Andric   for (auto &DS : Descriptors) {
129a7dea167SDimitry Andric     Scopes.emplace_back(std::move(DS));
130a7dea167SDimitry Andric   }
131a7dea167SDimitry Andric 
132a7dea167SDimitry Andric   // Set the function's code.
133a7dea167SDimitry Andric   Func->setCode(NextLocalOffset, std::move(Code), std::move(SrcMap),
13406c3fb27SDimitry Andric                 std::move(Scopes), FuncDecl->hasBody());
135bdd1243dSDimitry Andric   Func->setIsFullyCompiled(true);
136a7dea167SDimitry Andric   return Func;
137a7dea167SDimitry Andric }
138a7dea167SDimitry Andric 
139a7dea167SDimitry Andric Scope::Local ByteCodeEmitter::createLocal(Descriptor *D) {
140a7dea167SDimitry Andric   NextLocalOffset += sizeof(Block);
141a7dea167SDimitry Andric   unsigned Location = NextLocalOffset;
142a7dea167SDimitry Andric   NextLocalOffset += align(D->getAllocSize());
143a7dea167SDimitry Andric   return {Location, D};
144a7dea167SDimitry Andric }
145a7dea167SDimitry Andric 
146a7dea167SDimitry Andric void ByteCodeEmitter::emitLabel(LabelTy Label) {
147a7dea167SDimitry Andric   const size_t Target = Code.size();
148a7dea167SDimitry Andric   LabelOffsets.insert({Label, Target});
14906c3fb27SDimitry Andric 
15006c3fb27SDimitry Andric   if (auto It = LabelRelocs.find(Label);
15106c3fb27SDimitry Andric       It != LabelRelocs.end()) {
152a7dea167SDimitry Andric     for (unsigned Reloc : It->second) {
153a7dea167SDimitry Andric       using namespace llvm::support;
154a7dea167SDimitry Andric 
15506c3fb27SDimitry Andric       // Rewrite the operand of all jumps to this label.
156bdd1243dSDimitry Andric       void *Location = Code.data() + Reloc - align(sizeof(int32_t));
157bdd1243dSDimitry Andric       assert(aligned(Location));
158a7dea167SDimitry Andric       const int32_t Offset = Target - static_cast<int64_t>(Reloc);
1595f757f3fSDimitry Andric       endian::write<int32_t, llvm::endianness::native>(Location, Offset);
160a7dea167SDimitry Andric     }
161a7dea167SDimitry Andric     LabelRelocs.erase(It);
162a7dea167SDimitry Andric   }
163a7dea167SDimitry Andric }
164a7dea167SDimitry Andric 
165a7dea167SDimitry Andric int32_t ByteCodeEmitter::getOffset(LabelTy Label) {
166a7dea167SDimitry Andric   // Compute the PC offset which the jump is relative to.
167bdd1243dSDimitry Andric   const int64_t Position =
168bdd1243dSDimitry Andric       Code.size() + align(sizeof(Opcode)) + align(sizeof(int32_t));
169bdd1243dSDimitry Andric   assert(aligned(Position));
170a7dea167SDimitry Andric 
171a7dea167SDimitry Andric   // If target is known, compute jump offset.
17206c3fb27SDimitry Andric   if (auto It = LabelOffsets.find(Label);
17306c3fb27SDimitry Andric       It != LabelOffsets.end())
174a7dea167SDimitry Andric     return It->second - Position;
175a7dea167SDimitry Andric 
176a7dea167SDimitry Andric   // Otherwise, record relocation and return dummy offset.
177a7dea167SDimitry Andric   LabelRelocs[Label].push_back(Position);
178a7dea167SDimitry Andric   return 0ull;
179a7dea167SDimitry Andric }
180a7dea167SDimitry Andric 
181a7dea167SDimitry Andric /// Helper to write bytecode and bail out if 32-bit offsets become invalid.
182349cc55cSDimitry Andric /// Pointers will be automatically marshalled as 32-bit IDs.
183349cc55cSDimitry Andric template <typename T>
18406c3fb27SDimitry Andric static void emit(Program &P, std::vector<std::byte> &Code, const T &Val,
185bdd1243dSDimitry Andric                  bool &Success) {
186bdd1243dSDimitry Andric   size_t Size;
187bdd1243dSDimitry Andric 
188bdd1243dSDimitry Andric   if constexpr (std::is_pointer_v<T>)
189bdd1243dSDimitry Andric     Size = sizeof(uint32_t);
190bdd1243dSDimitry Andric   else
191bdd1243dSDimitry Andric     Size = sizeof(T);
192bdd1243dSDimitry Andric 
193a7dea167SDimitry Andric   if (Code.size() + Size > std::numeric_limits<unsigned>::max()) {
194a7dea167SDimitry Andric     Success = false;
195a7dea167SDimitry Andric     return;
196a7dea167SDimitry Andric   }
197349cc55cSDimitry Andric 
198bdd1243dSDimitry Andric   // Access must be aligned!
199bdd1243dSDimitry Andric   size_t ValPos = align(Code.size());
200bdd1243dSDimitry Andric   Size = align(Size);
201bdd1243dSDimitry Andric   assert(aligned(ValPos + Size));
202bdd1243dSDimitry Andric   Code.resize(ValPos + Size);
203349cc55cSDimitry Andric 
204bdd1243dSDimitry Andric   if constexpr (!std::is_pointer_v<T>) {
205bdd1243dSDimitry Andric     new (Code.data() + ValPos) T(Val);
206bdd1243dSDimitry Andric   } else {
207349cc55cSDimitry Andric     uint32_t ID = P.getOrCreateNativePointer(Val);
208bdd1243dSDimitry Andric     new (Code.data() + ValPos) uint32_t(ID);
209bdd1243dSDimitry Andric   }
210349cc55cSDimitry Andric }
211349cc55cSDimitry Andric 
2125f757f3fSDimitry Andric template <>
2135f757f3fSDimitry Andric void emit(Program &P, std::vector<std::byte> &Code, const Floating &Val,
2145f757f3fSDimitry Andric           bool &Success) {
2155f757f3fSDimitry Andric   size_t Size = Val.bytesToSerialize();
2165f757f3fSDimitry Andric 
2175f757f3fSDimitry Andric   if (Code.size() + Size > std::numeric_limits<unsigned>::max()) {
2185f757f3fSDimitry Andric     Success = false;
2195f757f3fSDimitry Andric     return;
2205f757f3fSDimitry Andric   }
2215f757f3fSDimitry Andric 
2225f757f3fSDimitry Andric   // Access must be aligned!
2235f757f3fSDimitry Andric   size_t ValPos = align(Code.size());
2245f757f3fSDimitry Andric   Size = align(Size);
2255f757f3fSDimitry Andric   assert(aligned(ValPos + Size));
2265f757f3fSDimitry Andric   Code.resize(ValPos + Size);
2275f757f3fSDimitry Andric 
2285f757f3fSDimitry Andric   Val.serialize(Code.data() + ValPos);
2295f757f3fSDimitry Andric }
2305f757f3fSDimitry Andric 
231349cc55cSDimitry Andric template <typename... Tys>
232349cc55cSDimitry Andric bool ByteCodeEmitter::emitOp(Opcode Op, const Tys &... Args, const SourceInfo &SI) {
233349cc55cSDimitry Andric   bool Success = true;
234a7dea167SDimitry Andric 
23506c3fb27SDimitry Andric   // The opcode is followed by arguments. The source info is
23606c3fb27SDimitry Andric   // attached to the address after the opcode.
237349cc55cSDimitry Andric   emit(P, Code, Op, Success);
238a7dea167SDimitry Andric   if (SI)
239a7dea167SDimitry Andric     SrcMap.emplace_back(Code.size(), SI);
240a7dea167SDimitry Andric 
24106c3fb27SDimitry Andric   // The initializer list forces the expression to be evaluated
24206c3fb27SDimitry Andric   // for each argument in the variadic template, in order.
243349cc55cSDimitry Andric   (void)std::initializer_list<int>{(emit(P, Code, Args, Success), 0)...};
244a7dea167SDimitry Andric 
245a7dea167SDimitry Andric   return Success;
246a7dea167SDimitry Andric }
247a7dea167SDimitry Andric 
248a7dea167SDimitry Andric bool ByteCodeEmitter::jumpTrue(const LabelTy &Label) {
249a7dea167SDimitry Andric   return emitJt(getOffset(Label), SourceInfo{});
250a7dea167SDimitry Andric }
251a7dea167SDimitry Andric 
252a7dea167SDimitry Andric bool ByteCodeEmitter::jumpFalse(const LabelTy &Label) {
253a7dea167SDimitry Andric   return emitJf(getOffset(Label), SourceInfo{});
254a7dea167SDimitry Andric }
255a7dea167SDimitry Andric 
256a7dea167SDimitry Andric bool ByteCodeEmitter::jump(const LabelTy &Label) {
257a7dea167SDimitry Andric   return emitJmp(getOffset(Label), SourceInfo{});
258a7dea167SDimitry Andric }
259a7dea167SDimitry Andric 
260a7dea167SDimitry Andric bool ByteCodeEmitter::fallthrough(const LabelTy &Label) {
261a7dea167SDimitry Andric   emitLabel(Label);
262a7dea167SDimitry Andric   return true;
263a7dea167SDimitry Andric }
264a7dea167SDimitry Andric 
265a7dea167SDimitry Andric //===----------------------------------------------------------------------===//
266a7dea167SDimitry Andric // Opcode emitters
267a7dea167SDimitry Andric //===----------------------------------------------------------------------===//
268a7dea167SDimitry Andric 
269a7dea167SDimitry Andric #define GET_LINK_IMPL
270a7dea167SDimitry Andric #include "Opcodes.inc"
271a7dea167SDimitry Andric #undef GET_LINK_IMPL
272