The comment already explains that this is only true for _some_ builtin
functions. This also brings the time it takes me to run the
builtin-functions.cpp test locally from 50s down to 47s.
This requires adding a new opcode for PointerToBoolean casts, since we
otherwise emit too many diagnostics. But that fixes an older problem
when casting weak pointers to bool.
FunctionDecl::getBuiltinID() is surprisingly slow and we tend to call it
quite a bit, especially when interpreting builtin functions. Caching the
BuiltinID here reduces the time I need to compile the
floating_comparison namespace from builtin-functions.cpp from 7.2s to
6.3s locally.
The PR https://github.com/llvm/llvm-project/pull/105996 broke taking the
address of a vector:
**compound-literal.c**
```C
typedef int v4i32 __attribute((vector_size(16)));
v4i32 *y = &(v4i32){1,2,3,4};
```
That because the current interpreter handle vector unary operator as a
fallback when the generic code path fail. but the new interpreter was
not. we need to handle `UO_AddrOf` in
`Compiler<Emitter>::VisitVectorUnaryOperator`.
Signed-off-by: yronglin <yronglin777@gmail.com>
Implement constexpr vector unary operators +, -, ~ and ! .
- Follow the current constant interpreter. All of our boolean operations
on vector types should be '-1' for the 'truth' type.
- Move the following functions from `Sema` to `ASTContext`, because we
used it in new interpreter.
```C++
QualType GetSignedVectorType(QualType V);
QualType GetSignedSizelessVectorType(QualType V);
```
---------
Signed-off-by: yronglin <yronglin777@gmail.com>
The new constant interpreter's `clang::interp::InterpState` contains
both `clang::interp::Context` and `clang::ASTContext`. So using `S.Ctx`
and `S.getCtx()` was a bit confusing. This PR rename `getCtx()` to
`getASTContext` to make things more clearer.
Signed-off-by: yronglin <yronglin777@gmail.com>
This reverts c79d1fa540390f6e37e1ea326153559eeadd0de6 and
125aa10b3d645bd26523a1bc321bb2e6b1cf04e1
Instead, use the previous approach but allow void-typed InitListExprs
with 0 initializers.
I started out by adding a new pointer type for blocks, and I was fully
prepared to compile their AST to bytecode and later call them.
... then I found out that the current interpreter doesn't support
calling blocks at all. So we reuse `Function` to support sources other
than `FunctionDecl`s and classify `BlockPointerType` as `PT_FnPtr`.
As per [P0533R9](https://wg21.link/P0533R9), the corresponding C++
`[c.math.fpclass]` standard library functions for the C macros are now
`constexpr`.
The only classification function that wasn't already `constexpr` was
`__builtin_signbit`.
The floating point comparison functions `__builtin_isgreater`,
`__builtin_isgreaterequal`, `__builtin_isless`, `__builtin_islessequal`,
`__builtin_islessgreater` and `__builtin_isunordered` are now
`constexpr`.
The C23 macro `iseqsig` is not currently supported because
`__bulitin_iseqsig` doesn't exist yet (and C++26 is still currently
based on C18).
This also allows them to be constant folded in C, matching the behaviour
of GCC.
... to current offset. This breaks other tests which this commit also
fixes. Namely, getIndex() should return the integer representation for
non-block pointers.
This happens for enum types with bool parent types. isBooleanType()
returns false for them however.
The previous version did the same thing by re-classifying the enum
integer type, but that breaks with forward-declared enums.
