llvm-project/clang/lib/AST/ByteCode/InterpStack.cpp
Timm Baeder 6ce13ae1c2
[clang][bytecode] Always track item types in InterpStack (#151088)
This has been a long-standing problem, but we didn't use to call the
destructors of items on the stack unless we explicitly `pop()` or
`discard()` them.

When interpretation was interrupted midway-through (because something
failed), we left `Pointer`s on the stack. Since all `Block`s track what
`Pointer`s point to them (via a doubly-linked list in the `Pointer`),
that meant we potentially leave deallocated pointers in that list. We
used to work around this by removing the `Pointer` from the list before
deallocating the block.

However, we now want to track pointers to global blocks as well, which
poses a problem since the blocks are never deallocated and thus those
pointers are always left dangling.

I've tried a few different approaches to fixing this but in the end I
just gave up on the idea of never knowing what items are in the stack.
We already have an `ItemTypes` vector that we use for debugging
assertions. This patch simply enables this vector unconditionally and
uses it in the abort case to properly `discard()` all elements from the
stack. That's a little sad IMO but I don't know of another way of
solving this problem.

As expected, this is a slight hit to compile times:
https://llvm-compile-time-tracker.com/compare.php?from=574d0a92060bf4808776b7a0239ffe91a092b15d&to=0317105f559093cfb909bfb01857a6b837991940&stat=instructions:u
2025-08-18 17:15:31 +02:00

133 lines
3.2 KiB
C++

//===--- InterpStack.cpp - Stack implementation for the VM ------*- 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
//
//===----------------------------------------------------------------------===//
#include "InterpStack.h"
#include "Boolean.h"
#include "FixedPoint.h"
#include "Floating.h"
#include "Integral.h"
#include "MemberPointer.h"
#include "Pointer.h"
#include <cassert>
#include <cstdlib>
using namespace clang;
using namespace clang::interp;
InterpStack::~InterpStack() {
if (Chunk && Chunk->Next)
std::free(Chunk->Next);
if (Chunk)
std::free(Chunk);
Chunk = nullptr;
StackSize = 0;
ItemTypes.clear();
}
// We keep the last chunk around to reuse.
void InterpStack::clear() {
for (PrimType Item : llvm::reverse(ItemTypes)) {
TYPE_SWITCH(Item, { this->discard<T>(); });
}
assert(ItemTypes.empty());
assert(empty());
}
void InterpStack::clearTo(size_t NewSize) {
if (NewSize == 0)
return clear();
if (NewSize == size())
return;
assert(NewSize <= size());
for (PrimType Item : llvm::reverse(ItemTypes)) {
TYPE_SWITCH(Item, { this->discard<T>(); });
if (size() == NewSize)
break;
}
// Note: discard() above already removed the types from ItemTypes.
assert(size() == NewSize);
}
void *InterpStack::grow(size_t Size) {
assert(Size < ChunkSize - sizeof(StackChunk) && "Object too large");
if (!Chunk || sizeof(StackChunk) + Chunk->size() + Size > ChunkSize) {
if (Chunk && Chunk->Next) {
Chunk = Chunk->Next;
} else {
StackChunk *Next = new (std::malloc(ChunkSize)) StackChunk(Chunk);
if (Chunk)
Chunk->Next = Next;
Chunk = Next;
}
}
auto *Object = reinterpret_cast<void *>(Chunk->End);
Chunk->End += Size;
StackSize += Size;
return Object;
}
void *InterpStack::peekData(size_t Size) const {
assert(Chunk && "Stack is empty!");
StackChunk *Ptr = Chunk;
while (Size > Ptr->size()) {
Size -= Ptr->size();
Ptr = Ptr->Prev;
assert(Ptr && "Offset too large");
}
return reinterpret_cast<void *>(Ptr->End - Size);
}
void InterpStack::shrink(size_t Size) {
assert(Chunk && "Chunk is empty!");
while (Size > Chunk->size()) {
Size -= Chunk->size();
if (Chunk->Next) {
std::free(Chunk->Next);
Chunk->Next = nullptr;
}
Chunk->End = Chunk->start();
Chunk = Chunk->Prev;
assert(Chunk && "Offset too large");
}
Chunk->End -= Size;
StackSize -= Size;
}
void InterpStack::dump() const {
llvm::errs() << "Items: " << ItemTypes.size() << ". Size: " << size() << '\n';
if (ItemTypes.empty())
return;
size_t Index = 0;
size_t Offset = 0;
// The type of the item on the top of the stack is inserted to the back
// of the vector, so the iteration has to happen backwards.
for (PrimType Item : llvm::reverse(ItemTypes)) {
Offset += align(primSize(Item));
llvm::errs() << Index << '/' << Offset << ": ";
TYPE_SWITCH(Item, {
const T &V = peek<T>(Offset);
llvm::errs() << V;
});
llvm::errs() << '\n';
++Index;
}
}