- //===- Allocator.h - Simple memory allocation abstraction -------*- C++ -*-===// 
- // 
- // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. 
- // See https://llvm.org/LICENSE.txt for license information. 
- // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception 
- // 
- //===----------------------------------------------------------------------===// 
- /// \file 
- /// 
- /// This file defines the BumpPtrAllocator interface. BumpPtrAllocator conforms 
- /// to the LLVM "Allocator" concept and is similar to MallocAllocator, but 
- /// objects cannot be deallocated. Their lifetime is tied to the lifetime of the 
- /// allocator. 
- /// 
- //===----------------------------------------------------------------------===// 
-   
- #ifndef LLVM_SUPPORT_ALLOCATOR_H 
- #define LLVM_SUPPORT_ALLOCATOR_H 
-   
- #include "llvm/ADT/SmallVector.h" 
- #include "llvm/Support/Alignment.h" 
- #include "llvm/Support/AllocatorBase.h" 
- #include "llvm/Support/Compiler.h" 
- #include "llvm/Support/MathExtras.h" 
- #include <algorithm> 
- #include <cassert> 
- #include <cstddef> 
- #include <cstdint> 
- #include <iterator> 
- #include <optional> 
- #include <utility> 
-   
- namespace llvm { 
-   
- namespace detail { 
-   
- // We call out to an external function to actually print the message as the 
- // printing code uses Allocator.h in its implementation. 
- void printBumpPtrAllocatorStats(unsigned NumSlabs, size_t BytesAllocated, 
-                                 size_t TotalMemory); 
-   
- } // end namespace detail 
-   
- /// Allocate memory in an ever growing pool, as if by bump-pointer. 
- /// 
- /// This isn't strictly a bump-pointer allocator as it uses backing slabs of 
- /// memory rather than relying on a boundless contiguous heap. However, it has 
- /// bump-pointer semantics in that it is a monotonically growing pool of memory 
- /// where every allocation is found by merely allocating the next N bytes in 
- /// the slab, or the next N bytes in the next slab. 
- /// 
- /// Note that this also has a threshold for forcing allocations above a certain 
- /// size into their own slab. 
- /// 
- /// The BumpPtrAllocatorImpl template defaults to using a MallocAllocator 
- /// object, which wraps malloc, to allocate memory, but it can be changed to 
- /// use a custom allocator. 
- /// 
- /// The GrowthDelay specifies after how many allocated slabs the allocator 
- /// increases the size of the slabs. 
- template <typename AllocatorT = MallocAllocator, size_t SlabSize = 4096, 
-           size_t SizeThreshold = SlabSize, size_t GrowthDelay = 128> 
- class BumpPtrAllocatorImpl 
-     : public AllocatorBase<BumpPtrAllocatorImpl<AllocatorT, SlabSize, 
-                                                 SizeThreshold, GrowthDelay>>, 
-       private detail::AllocatorHolder<AllocatorT> { 
-   using AllocTy = detail::AllocatorHolder<AllocatorT>; 
-   
- public: 
-   static_assert(SizeThreshold <= SlabSize, 
-                 "The SizeThreshold must be at most the SlabSize to ensure " 
-                 "that objects larger than a slab go into their own memory " 
-                 "allocation."); 
-   static_assert(GrowthDelay > 0, 
-                 "GrowthDelay must be at least 1 which already increases the" 
-                 "slab size after each allocated slab."); 
-   
-   BumpPtrAllocatorImpl() = default; 
-   
-   template <typename T> 
-   BumpPtrAllocatorImpl(T &&Allocator) 
-       : AllocTy(std::forward<T &&>(Allocator)) {} 
-   
-   // Manually implement a move constructor as we must clear the old allocator's 
-   // slabs as a matter of correctness. 
-   BumpPtrAllocatorImpl(BumpPtrAllocatorImpl &&Old) 
-       : AllocTy(std::move(Old.getAllocator())), CurPtr(Old.CurPtr), 
-         End(Old.End), Slabs(std::move(Old.Slabs)), 
-         CustomSizedSlabs(std::move(Old.CustomSizedSlabs)), 
-         BytesAllocated(Old.BytesAllocated), RedZoneSize(Old.RedZoneSize) { 
-     Old.CurPtr = Old.End = nullptr; 
-     Old.BytesAllocated = 0; 
-     Old.Slabs.clear(); 
-     Old.CustomSizedSlabs.clear(); 
-   } 
-   
-   ~BumpPtrAllocatorImpl() { 
-     DeallocateSlabs(Slabs.begin(), Slabs.end()); 
-     DeallocateCustomSizedSlabs(); 
-   } 
-   
-   BumpPtrAllocatorImpl &operator=(BumpPtrAllocatorImpl &&RHS) { 
-     DeallocateSlabs(Slabs.begin(), Slabs.end()); 
-     DeallocateCustomSizedSlabs(); 
-   
-     CurPtr = RHS.CurPtr; 
-     End = RHS.End; 
-     BytesAllocated = RHS.BytesAllocated; 
-     RedZoneSize = RHS.RedZoneSize; 
-     Slabs = std::move(RHS.Slabs); 
-     CustomSizedSlabs = std::move(RHS.CustomSizedSlabs); 
-     AllocTy::operator=(std::move(RHS.getAllocator())); 
-   
-     RHS.CurPtr = RHS.End = nullptr; 
-     RHS.BytesAllocated = 0; 
-     RHS.Slabs.clear(); 
-     RHS.CustomSizedSlabs.clear(); 
-     return *this; 
-   } 
-   
-   /// Deallocate all but the current slab and reset the current pointer 
-   /// to the beginning of it, freeing all memory allocated so far. 
-   void Reset() { 
-     // Deallocate all but the first slab, and deallocate all custom-sized slabs. 
-     DeallocateCustomSizedSlabs(); 
-     CustomSizedSlabs.clear(); 
-   
-     if (Slabs.empty()) 
-       return; 
-   
-     // Reset the state. 
-     BytesAllocated = 0; 
-     CurPtr = (char *)Slabs.front(); 
-     End = CurPtr + SlabSize; 
-   
-     __asan_poison_memory_region(*Slabs.begin(), computeSlabSize(0)); 
-     DeallocateSlabs(std::next(Slabs.begin()), Slabs.end()); 
-     Slabs.erase(std::next(Slabs.begin()), Slabs.end()); 
-   } 
-   
-   /// Allocate space at the specified alignment. 
-   // This method is *not* marked noalias, because 
-   // SpecificBumpPtrAllocator::DestroyAll() loops over all allocations, and 
-   // that loop is not based on the Allocate() return value. 
-   // 
-   // Allocate(0, N) is valid, it returns a non-null pointer (which should not 
-   // be dereferenced). 
-   LLVM_ATTRIBUTE_RETURNS_NONNULL void *Allocate(size_t Size, Align Alignment) { 
-     // Keep track of how many bytes we've allocated. 
-     BytesAllocated += Size; 
-   
-     size_t Adjustment = offsetToAlignedAddr(CurPtr, Alignment); 
-     assert(Adjustment + Size >= Size && "Adjustment + Size must not overflow"); 
-   
-     size_t SizeToAllocate = Size; 
- #if LLVM_ADDRESS_SANITIZER_BUILD 
-     // Add trailing bytes as a "red zone" under ASan. 
-     SizeToAllocate += RedZoneSize; 
- #endif 
-   
-     // Check if we have enough space. 
-     if (Adjustment + SizeToAllocate <= size_t(End - CurPtr) 
-         // We can't return nullptr even for a zero-sized allocation! 
-         && CurPtr != nullptr) { 
-       char *AlignedPtr = CurPtr + Adjustment; 
-       CurPtr = AlignedPtr + SizeToAllocate; 
-       // Update the allocation point of this memory block in MemorySanitizer. 
-       // Without this, MemorySanitizer messages for values originated from here 
-       // will point to the allocation of the entire slab. 
-       __msan_allocated_memory(AlignedPtr, Size); 
-       // Similarly, tell ASan about this space. 
-       __asan_unpoison_memory_region(AlignedPtr, Size); 
-       return AlignedPtr; 
-     } 
-   
-     // If Size is really big, allocate a separate slab for it. 
-     size_t PaddedSize = SizeToAllocate + Alignment.value() - 1; 
-     if (PaddedSize > SizeThreshold) { 
-       void *NewSlab = 
-           this->getAllocator().Allocate(PaddedSize, alignof(std::max_align_t)); 
-       // We own the new slab and don't want anyone reading anyting other than 
-       // pieces returned from this method.  So poison the whole slab. 
-       __asan_poison_memory_region(NewSlab, PaddedSize); 
-       CustomSizedSlabs.push_back(std::make_pair(NewSlab, PaddedSize)); 
-   
-       uintptr_t AlignedAddr = alignAddr(NewSlab, Alignment); 
-       assert(AlignedAddr + Size <= (uintptr_t)NewSlab + PaddedSize); 
-       char *AlignedPtr = (char*)AlignedAddr; 
-       __msan_allocated_memory(AlignedPtr, Size); 
-       __asan_unpoison_memory_region(AlignedPtr, Size); 
-       return AlignedPtr; 
-     } 
-   
-     // Otherwise, start a new slab and try again. 
-     StartNewSlab(); 
-     uintptr_t AlignedAddr = alignAddr(CurPtr, Alignment); 
-     assert(AlignedAddr + SizeToAllocate <= (uintptr_t)End && 
-            "Unable to allocate memory!"); 
-     char *AlignedPtr = (char*)AlignedAddr; 
-     CurPtr = AlignedPtr + SizeToAllocate; 
-     __msan_allocated_memory(AlignedPtr, Size); 
-     __asan_unpoison_memory_region(AlignedPtr, Size); 
-     return AlignedPtr; 
-   } 
-   
-   inline LLVM_ATTRIBUTE_RETURNS_NONNULL void * 
-   Allocate(size_t Size, size_t Alignment) { 
-     assert(Alignment > 0 && "0-byte alignment is not allowed. Use 1 instead."); 
-     return Allocate(Size, Align(Alignment)); 
-   } 
-   
-   // Pull in base class overloads. 
-   using AllocatorBase<BumpPtrAllocatorImpl>::Allocate; 
-   
-   // Bump pointer allocators are expected to never free their storage; and 
-   // clients expect pointers to remain valid for non-dereferencing uses even 
-   // after deallocation. 
-   void Deallocate(const void *Ptr, size_t Size, size_t /*Alignment*/) { 
-     __asan_poison_memory_region(Ptr, Size); 
-   } 
-   
-   // Pull in base class overloads. 
-   using AllocatorBase<BumpPtrAllocatorImpl>::Deallocate; 
-   
-   size_t GetNumSlabs() const { return Slabs.size() + CustomSizedSlabs.size(); } 
-   
-   /// \return An index uniquely and reproducibly identifying 
-   /// an input pointer \p Ptr in the given allocator. 
-   /// The returned value is negative iff the object is inside a custom-size 
-   /// slab. 
-   /// Returns an empty optional if the pointer is not found in the allocator. 
-   std::optional<int64_t> identifyObject(const void *Ptr) { 
-     const char *P = static_cast<const char *>(Ptr); 
-     int64_t InSlabIdx = 0; 
-     for (size_t Idx = 0, E = Slabs.size(); Idx < E; Idx++) { 
-       const char *S = static_cast<const char *>(Slabs[Idx]); 
-       if (P >= S && P < S + computeSlabSize(Idx)) 
-         return InSlabIdx + static_cast<int64_t>(P - S); 
-       InSlabIdx += static_cast<int64_t>(computeSlabSize(Idx)); 
-     } 
-   
-     // Use negative index to denote custom sized slabs. 
-     int64_t InCustomSizedSlabIdx = -1; 
-     for (size_t Idx = 0, E = CustomSizedSlabs.size(); Idx < E; Idx++) { 
-       const char *S = static_cast<const char *>(CustomSizedSlabs[Idx].first); 
-       size_t Size = CustomSizedSlabs[Idx].second; 
-       if (P >= S && P < S + Size) 
-         return InCustomSizedSlabIdx - static_cast<int64_t>(P - S); 
-       InCustomSizedSlabIdx -= static_cast<int64_t>(Size); 
-     } 
-     return std::nullopt; 
-   } 
-   
-   /// A wrapper around identifyObject that additionally asserts that 
-   /// the object is indeed within the allocator. 
-   /// \return An index uniquely and reproducibly identifying 
-   /// an input pointer \p Ptr in the given allocator. 
-   int64_t identifyKnownObject(const void *Ptr) { 
-     std::optional<int64_t> Out = identifyObject(Ptr); 
-     assert(Out && "Wrong allocator used"); 
-     return *Out; 
-   } 
-   
-   /// A wrapper around identifyKnownObject. Accepts type information 
-   /// about the object and produces a smaller identifier by relying on 
-   /// the alignment information. Note that sub-classes may have different 
-   /// alignment, so the most base class should be passed as template parameter 
-   /// in order to obtain correct results. For that reason automatic template 
-   /// parameter deduction is disabled. 
-   /// \return An index uniquely and reproducibly identifying 
-   /// an input pointer \p Ptr in the given allocator. This identifier is 
-   /// different from the ones produced by identifyObject and 
-   /// identifyAlignedObject. 
-   template <typename T> 
-   int64_t identifyKnownAlignedObject(const void *Ptr) { 
-     int64_t Out = identifyKnownObject(Ptr); 
-     assert(Out % alignof(T) == 0 && "Wrong alignment information"); 
-     return Out / alignof(T); 
-   } 
-   
-   size_t getTotalMemory() const { 
-     size_t TotalMemory = 0; 
-     for (auto I = Slabs.begin(), E = Slabs.end(); I != E; ++I) 
-       TotalMemory += computeSlabSize(std::distance(Slabs.begin(), I)); 
-     for (const auto &PtrAndSize : CustomSizedSlabs) 
-       TotalMemory += PtrAndSize.second; 
-     return TotalMemory; 
-   } 
-   
-   size_t getBytesAllocated() const { return BytesAllocated; } 
-   
-   void setRedZoneSize(size_t NewSize) { 
-     RedZoneSize = NewSize; 
-   } 
-   
-   void PrintStats() const { 
-     detail::printBumpPtrAllocatorStats(Slabs.size(), BytesAllocated, 
-                                        getTotalMemory()); 
-   } 
-   
- private: 
-   /// The current pointer into the current slab. 
-   /// 
-   /// This points to the next free byte in the slab. 
-   char *CurPtr = nullptr; 
-   
-   /// The end of the current slab. 
-   char *End = nullptr; 
-   
-   /// The slabs allocated so far. 
-   SmallVector<void *, 4> Slabs; 
-   
-   /// Custom-sized slabs allocated for too-large allocation requests. 
-   SmallVector<std::pair<void *, size_t>, 0> CustomSizedSlabs; 
-   
-   /// How many bytes we've allocated. 
-   /// 
-   /// Used so that we can compute how much space was wasted. 
-   size_t BytesAllocated = 0; 
-   
-   /// The number of bytes to put between allocations when running under 
-   /// a sanitizer. 
-   size_t RedZoneSize = 1; 
-   
-   static size_t computeSlabSize(unsigned SlabIdx) { 
-     // Scale the actual allocated slab size based on the number of slabs 
-     // allocated. Every GrowthDelay slabs allocated, we double 
-     // the allocated size to reduce allocation frequency, but saturate at 
-     // multiplying the slab size by 2^30. 
-     return SlabSize * 
-            ((size_t)1 << std::min<size_t>(30, SlabIdx / GrowthDelay)); 
-   } 
-   
-   /// Allocate a new slab and move the bump pointers over into the new 
-   /// slab, modifying CurPtr and End. 
-   void StartNewSlab() { 
-     size_t AllocatedSlabSize = computeSlabSize(Slabs.size()); 
-   
-     void *NewSlab = this->getAllocator().Allocate(AllocatedSlabSize, 
-                                                   alignof(std::max_align_t)); 
-     // We own the new slab and don't want anyone reading anything other than 
-     // pieces returned from this method.  So poison the whole slab. 
-     __asan_poison_memory_region(NewSlab, AllocatedSlabSize); 
-   
-     Slabs.push_back(NewSlab); 
-     CurPtr = (char *)(NewSlab); 
-     End = ((char *)NewSlab) + AllocatedSlabSize; 
-   } 
-   
-   /// Deallocate a sequence of slabs. 
-   void DeallocateSlabs(SmallVectorImpl<void *>::iterator I, 
-                        SmallVectorImpl<void *>::iterator E) { 
-     for (; I != E; ++I) { 
-       size_t AllocatedSlabSize = 
-           computeSlabSize(std::distance(Slabs.begin(), I)); 
-       this->getAllocator().Deallocate(*I, AllocatedSlabSize, 
-                                       alignof(std::max_align_t)); 
-     } 
-   } 
-   
-   /// Deallocate all memory for custom sized slabs. 
-   void DeallocateCustomSizedSlabs() { 
-     for (auto &PtrAndSize : CustomSizedSlabs) { 
-       void *Ptr = PtrAndSize.first; 
-       size_t Size = PtrAndSize.second; 
-       this->getAllocator().Deallocate(Ptr, Size, alignof(std::max_align_t)); 
-     } 
-   } 
-   
-   template <typename T> friend class SpecificBumpPtrAllocator; 
- }; 
-   
- /// The standard BumpPtrAllocator which just uses the default template 
- /// parameters. 
- typedef BumpPtrAllocatorImpl<> BumpPtrAllocator; 
-   
- /// A BumpPtrAllocator that allows only elements of a specific type to be 
- /// allocated. 
- /// 
- /// This allows calling the destructor in DestroyAll() and when the allocator is 
- /// destroyed. 
- template <typename T> class SpecificBumpPtrAllocator { 
-   BumpPtrAllocator Allocator; 
-   
- public: 
-   SpecificBumpPtrAllocator() { 
-     // Because SpecificBumpPtrAllocator walks the memory to call destructors, 
-     // it can't have red zones between allocations. 
-     Allocator.setRedZoneSize(0); 
-   } 
-   SpecificBumpPtrAllocator(SpecificBumpPtrAllocator &&Old) 
-       : Allocator(std::move(Old.Allocator)) {} 
-   ~SpecificBumpPtrAllocator() { DestroyAll(); } 
-   
-   SpecificBumpPtrAllocator &operator=(SpecificBumpPtrAllocator &&RHS) { 
-     Allocator = std::move(RHS.Allocator); 
-     return *this; 
-   } 
-   
-   /// Call the destructor of each allocated object and deallocate all but the 
-   /// current slab and reset the current pointer to the beginning of it, freeing 
-   /// all memory allocated so far. 
-   void DestroyAll() { 
-     auto DestroyElements = [](char *Begin, char *End) { 
-       assert(Begin == (char *)alignAddr(Begin, Align::Of<T>())); 
-       for (char *Ptr = Begin; Ptr + sizeof(T) <= End; Ptr += sizeof(T)) 
-         reinterpret_cast<T *>(Ptr)->~T(); 
-     }; 
-   
-     for (auto I = Allocator.Slabs.begin(), E = Allocator.Slabs.end(); I != E; 
-          ++I) { 
-       size_t AllocatedSlabSize = BumpPtrAllocator::computeSlabSize( 
-           std::distance(Allocator.Slabs.begin(), I)); 
-       char *Begin = (char *)alignAddr(*I, Align::Of<T>()); 
-       char *End = *I == Allocator.Slabs.back() ? Allocator.CurPtr 
-                                                : (char *)*I + AllocatedSlabSize; 
-   
-       DestroyElements(Begin, End); 
-     } 
-   
-     for (auto &PtrAndSize : Allocator.CustomSizedSlabs) { 
-       void *Ptr = PtrAndSize.first; 
-       size_t Size = PtrAndSize.second; 
-       DestroyElements((char *)alignAddr(Ptr, Align::Of<T>()), 
-                       (char *)Ptr + Size); 
-     } 
-   
-     Allocator.Reset(); 
-   } 
-   
-   /// Allocate space for an array of objects without constructing them. 
-   T *Allocate(size_t num = 1) { return Allocator.Allocate<T>(num); } 
- }; 
-   
- } // end namespace llvm 
-   
- template <typename AllocatorT, size_t SlabSize, size_t SizeThreshold, 
-           size_t GrowthDelay> 
- void * 
- operator new(size_t Size, 
-              llvm::BumpPtrAllocatorImpl<AllocatorT, SlabSize, SizeThreshold, 
-                                         GrowthDelay> &Allocator) { 
-   return Allocator.Allocate(Size, std::min((size_t)llvm::NextPowerOf2(Size), 
-                                            alignof(std::max_align_t))); 
- } 
-   
- template <typename AllocatorT, size_t SlabSize, size_t SizeThreshold, 
-           size_t GrowthDelay> 
- void operator delete(void *, 
-                      llvm::BumpPtrAllocatorImpl<AllocatorT, SlabSize, 
-                                                 SizeThreshold, GrowthDelay> &) { 
- } 
-   
- #endif // LLVM_SUPPORT_ALLOCATOR_H 
-