Constant Interpreter¶

Introduction ¶

The bytecode interpreter aims to replace the existing AST traversal-based evaluator in Clang, improving performance on constructs which are executed inefficiently by the evaluator. The interpreter is activated by passing -fexperimental-new-constant-interpreter to clang.

Since Clang 23, the bytecode interpreter can also be enabled by default by passing -DCLANG_USE_EXPERIMENTAL_CONST_INTERP=ON to cmake. In that case, it can be deactivated again via -fno-experimental-new-constant-interpreter.

Bytecode Compilation ¶

Bytecode compilation is handled in Compiler.h for statements and for expressions. The compiler has two different backends: one to generate bytecode for functions (ByteCodeEmitter) and one to directly evaluate expressions as they are compiled, without generating bytecode (EvalEmitter). All functions are compiled to bytecode, while toplevel expressions used in constant contexts are directly evaluated since the bytecode would never be reused. This mechanism aims to pave the way towards replacing the evaluator, improving its performance on functions and loops, while being just as fast on single-use toplevel expressions.

The interpreter relies on stack-based, strongly-typed opcodes. The glue logic between the code generator, along with the enumeration and description of opcodes, can be found in Opcodes.td. The opcodes are implemented as generic template methods in Interp.h and instantiated with the relevant primitive types by the interpreter loop or by the evaluating emitter.

Primitive Types ¶

PT_{U|S}int{8|16|32|64}

Signed or unsigned integers of a specific bit width. 1-byte types are also 1 byte in size. Sizes >= 16 bits are implemented using the Integral class. they take up 24 bytes each since they need to be able to represent a pointer that has been casted to an integer.
PT_IntAP{S}

Signed or unsigned integers of an arbitrary, but fixed width used to implement integral types which are required by the target, but are not supported by the host. Under the hood, they rely on APInt. The Integral specialisation for these types is required by opcodes to share an implementation with fixed integrals.
PT_Bool

Representation for boolean types, essentially a 1-bit unsigned Integral.
PT_Float

Arbitrary, but fixed precision floating point numbers. Could be specialised in the future similarly to integers in order to improve floating point performance.
PT_Ptr

Pointer type, defined in "Pointer.h". The most common type of pointer is a “BlockPointer”, which points to an interp::Block. But other pointer types exist, such as typeid pointers or integral pointers.
PT_MemberPtr

Member pointer type, can also be a null member pointer. Defined in "MemberPointer.h"

Composite types ¶

The interpreter distinguishes two kinds of composite types: arrays and records (structs and classes). Unions are represented as records, except at most a single field can be marked as active. The contents of inactive fields are kept until they are reactivated and overwritten. Complex numbers (_Complex) and vectors (__attribute((vector_size(16)))) are treated as arrays.

Bytecode Execution ¶

Bytecode is executed using a stack-based interpreter. The execution context consists of an InterpStack, along with a chain of InterpFrame objects storing the call frames. Frames are built by call instructions and destroyed by return instructions. They perform one allocation to reserve space for all locals in a single block. These objects store all the required information to emit stack traces whenever evaluation fails.

Memory Organisation ¶

Memory management in the interpreter relies on 3 data structures: Block objects which store the data and associated inline metadata, Pointer objects which refer to or into blocks, and Descriptor structures which describe blocks and subobjects nested inside blocks.

Blocks ¶

Blocks contain data interleaved with metadata. They are allocated either statically in the code generator (globals, static members, dummy parameter values etc.) or dynamically in the interpreter, when creating the frame containing the local variables of a function. Blocks are associated with a descriptor that characterises the entire allocation, along with a few additional attributes:

IsStatic indicates whether the block has static duration in the interpreter, i.e. it is not a local in a frame.
DeclID identifies each global declaration (it is set to an invalid and irrelevant value for locals) in order to prevent illegal writes and reads involving globals and temporaries with static storage duration.

Static blocks are never deallocated, but local ones might be deallocated even when there are live pointers to them. Pointers are only valid as long as the blocks they point to are valid, so a block with pointers to it whose lifetime ends is kept alive until all pointers to it go out of scope. Since the frame is destroyed on function exit, such blocks are turned into a DeadBlock and copied to storage managed by the interpreter itself, not the frame. Reads and writes to these blocks are illegal and cause an appropriate diagnostic to be emitted. When the last pointer goes out of scope, dead blocks are also deallocated.

The lifetime of blocks is managed through 2 methods stored in the descriptor of the block:

CtorFn: initializes the metadata which is stored in the block, alongside actual data. Invokes the default constructors of objects which are not trivial (Pointer, Floating, etc.)
DtorFn: invokes the destructors of non-trivial objects.

Blocks track all the pointers into them through an intrusive doubly-linked list, required to adjust and invalidate all pointers when transforming a block into a dead block. If the lifetime of an object ends, all pointers to it are invalidated, emitting the appropriate diagnostics when dereferenced.

Records are laid out identically to arrays of composites: each field and base class is preceded by an inline descriptor. The InlineDescriptor saves information about the initialization state, constness, mutability, lifetime, etc. of a field.

Consider this struct:

struct S {
    char c;
};
constexpr S s{12};

When allocating space for s, we allocate 24 bytes (not counting the Descriptor instances we created for the Record and the field). The field c needs 8 bytes, since we align to pointer size (this example uses a 64 bit system). The InlineDescriptor preceding the field data uses up the remaining 16 bytes.

Inline descriptors are filled in by the CtorFn of blocks, which leaves storage in an uninitialised, but valid state.

Descriptors ¶

Descriptors are generated at bytecode compilation time and contain information required to determine if a particular memory access is allowed in constexpr. They also carry all the information required to emit a diagnostic involving a memory access, such as the declaration which originates the block. Currently there is a single kind of descriptor encoding information for all block types.

Pointers ¶

Pointers, implemented in Pointer.h are represented as a tagged union.

BlockPointer: used to reference memory allocated and managed by the interpreter, being the only pointer kind which allows dereferencing in the interpreter

TypeIDPointer: tracks information for the opaque type returned by typeid

IntegralPointer: a pointer formed from an integer, think (int*)123.

FunctionPointer: a pointer to a function.

Besides the previously mentioned union, a number of other pointer-like types have their own type:

FunctionPointer tracks functions.

MemberPointer tracks C++ object members

BlockPointer ¶

Block pointers track a Pointee, the block to which they point, along with a Base and an Offset. The base identifies the innermost field, while the offset points to an array element relative to the base (including one-past-end pointers). The offset identifies the array element or field which is referenced, while the base points to the outer object or array which contains the field. These two fields allow all pointers to be uniquely identified, disambiguated and characterised.

As an example, consider the following structure:

struct A {
    struct B {
        int x;
        int y;
    } b;
    struct C {
        int a;
        int b;
    } c[2];
    int z;
};
constexpr A a;

On the target, &a and &a.b.x are equal. So are &a.c[0] and &a.c[0].a. In the interpreter, all these pointers must be distinguished since they are all allowed to address a distinct range of memory.

In the interpreter, the object would require 240 bytes of storage and would have its fields interleaved with metadata. The pointers which can be derived to the object are illustrated in the following diagram:

    0   16  32  40  56  64  80  96  112 120 136 144 160 176 184 200 208 224 240
+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+
+ B | D | D | x | D | y | D | D | D | a | D | b | D | D | a | D | b | D | z |
+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+
    ^   ^   ^       ^       ^   ^   ^       ^       ^   ^       ^       ^
    |   |   |       |       |   |   |   &a.c[0].b   |   |   &a.c[1].b   |
    a   |&a.b.x   &a.y    &a.c  |&a.c[0].a          |&a.c[1].a          |
      &a.b                   &a.c[0]            &a.c[1]               &a.z

The Base offset of all pointers points to the start of a field or an array and is preceded by an inline descriptor (unless Base is zero, pointing to the root). All the relevant attributes can be read from either the inline descriptor or the descriptor of the block.

Array elements are identified by the Offset field of pointers, pointing to past the inline descriptors for composites and before the actual data in the case of primitive arrays. The Offset points to the offset where primitives can be read from. As an example, a.c + 1 would have the same base as a.c since it is an element of a.c, but its offset would point to &a.c[1]. The array-to-pointer decay operation adjusts a pointer to an array (where the offset is equal to the base) to a pointer to the first element.

TypeInfoPointer ¶

TypeInfoPointer tracks two types: the type assigned to std::type_info and the type which was passed to typeinfo. It is part of the tagged union in Pointer.

Interpretation ¶

After bytecode has been generated (or not, for expressions), the bytecode is then interpreted. The bytecode is stack-based and uses InterpStack to allocate memory for the produced values.

Here is an example function:

constexpr int add(int a, int b) {
  return a + b;
}
static_assert(add(1, 2) == 3);

Which generates the following bytecode (this can be produced via interp::Function::dump()):

add 0x7cb97f7e2000
[...]
   GetParamSint32    0
  GetParamSint32    1
  AddSint32
  RetSint32
  NoRet

As you can see, all instructions here are type-aware. We’re first pushing both parameter values to the stack. Then the Add opcode will add them up and push the result to the stack, which will be returned via the Ret opcode.

Debugging ¶

Here are a few hints when working on the bytecode interpreter:

Setting a breakpoint on CCEDiag and FFDiag will stop the debugger when a diagnostic is emitted.
If you want to see the bytecode of a function call dump() on the interp::Function.
Additionally to the last point, interp::Function::dump(CodePtr) exists, which you can pass the current OpPC and it will then show you where in the bytecode that opcode is.
interp::Pointer has an operator<< that prints useful information. Try it e.g. via llvm::errs() << Ptr << '\n';.
Printing APValue instances also works via APValue::dump().
If you want to see everything that’s being evaluated, add debugging output to the evaluate* functions in Context.cpp.