Commit Graph
14 Commits
Author SHA1 Message Date
Boaz Brickner 508a88e2a9 C++ inteop: Set type source info for a generated C++ thunk function (#6049)
This prevents a null pointer access crash when generating a thunk with
an automatically deduced trivial return type.
In this case, Clang calls `Sema::DeduceFunctionTypeFromReturnExpr()`
which calls `Sema::getReturnTypeLoc()`, which requires this information.

Part of #5514.
2025-09-12 07:11:36 +00:00
Chandler Carruth 0691d4827c Update LLVM (#6022)
Notably, this updates past a major AST refactoring and tries to apply
those changes across the toolchain.
2025-09-06 08:36:10 +00:00
Richard Smith db0a00d713 Fix double-destruction of temporaries. (#6010)
Attach the cleanup to the `Temporary` instruction instead of to the
`TemporaryStorage` instruction. We create `TemporaryStorage`
instructions speculatively when creating an initializing expression, and
may overwrite those instructions with other instructions if it turns out
that a temporary is not required. Instead, wait until we finalize the
temporary and create a `Temporary` instruction to register the cleanup.
2025-09-04 19:19:59 +00:00
Richard Smith 3533668186 Support for building thunks for C++ constructors. (#5977) 2025-08-26 21:49:11 +00:00
Richard Smith d37f1ae6b5 Add return value support to C++ thunks. (#5976)
Based on #5948. A couple of tricky parts:

* When generating the C++ side of the thunk, we are given a pointer to
the location to emplace the return value. The only mechanism C++
provides to perform this emplacement is using placement `operator new`,
which requires a library function in the `<new>` header. We handle this
by declaring that library function ourselves, and rely on Clang not
actually needing a definition for it (which the standard library owns).

* On the Carbon side of the thunk, we want to form an initializing
expression as the result of the call. We don't have a way of expressing
in SemIR that an initializing expression performs its initialization by
storing through a pointer, so this PR adds a new initializing
instruction, `InPlaceInit`, to model an initialization that's performed
opaquely in-place.
2025-08-26 02:23:38 +00:00
Richard Smith 8c9080801c Support for building thunks for C++ methods. (#5972)
Also tweak how we import C++ methods to properly handle C++23's explicit
object parameters.
2025-08-25 22:33:51 +00:00
Richard Smith ddafbc9331 Use direct passing for 32- and 64-bit unsigned integers. (#5980)
Previously we only avoided creating a thunk for signed integers. But the
same logic also applies to the unsigned 32-bit and 64-bit types.
2025-08-25 21:49:03 +00:00
Boaz Brickner 5b328da0aa Avoid creating a vector of bools to mark which C++ thunk parameter types were modified (#5941)
Instead, check which types were modified.
This simplifies the logic and could make it easier to add return value
support.

Part of #5514.
2025-08-11 15:01:51 +00:00
Boaz Brickner fd3eb136af Use llvm::zip() to iterate over callee_function_params, thunk_function_params and callee_arg_ids (#5940) 2025-08-11 14:57:42 +00:00
Richard Smith e0de9ddf05 Implement initialization for C++ thunk parameters. (#5938)
When initializing a C++ thunk parameter:

* If we have an initializing expression, materialize a temporary and
pass its address.
* If we have a reference expression, pass its address directly.
* If we have a value expression with a pointer value representation,
pass the pointer.
* Otherwise, create a new temporary and initialize it with a copy of the
argument, and pass its address.
2025-08-08 23:38:47 +00:00
Boaz Brickner 9f108bad6e Rename cpp_ast to clang_ast_unit (#5926)
Followup of #5924.
2025-08-08 08:11:49 +00:00
Boaz Brickner 52ed26235d Add a flag to dump the C++ AST (#5918)
Use it to dump the AST that includes a generated C++ thunk.
Based on #5917.

Also added printing of the full actual text when check fails to make
debugging easier.
Changed line replacement to allow removing complete lines.

Part of #5514.
2025-08-07 17:09:30 +00:00
Boaz Brickner 0f171611d4 Create clang::MangleContext once per file instead of once per C++ thunk (#5924)
Part of #5514.
2025-08-07 14:50:46 +00:00
Boaz Brickner 3c9d267388 Generate and use a C++ thunk to call non simple ABI C++ functions (#5850)
When the C++ function has a parameter that is not a pointer and not a
signed integer of 32 or 64 bits, generate a thunk.

Terminology:
* Callee function: The C++ function we actually want to call.
* Thunk function: The C++ function we generated that calls the callee
function.
* A simple ABI type, for now, is one of:
  * A pointer
  * signed integer with 32 bits
  * signed integer with 64 bits

The thunk function is marked `always_inline` and uses the `asm`
attribute to set its mangled to the callee function mangled name
suffixed with `".carbon_thunk"`.

When importing a C++ function, we decide whether calling it requires a
thunk and if so we generate it and import it as well, which is currently
a recursive call.

When calling the thunk function, we initialize a temporary storage for
each non simple ABI parameter type and take its address. This can be
optimized when the variable is already in storage.

Not supported yet:
* Functions with non void return values.
* Member methods.

Moved unsigned int param test from `arithmetic_types_direct.carbon` to
`arithmetic_types_bridged.carbon`, since only signed integers aren't
bridged using a thunk.

C++ Interop Demo:

```c++
// hello_world.h

struct S {
  S() {}
  S(const S&) { x = 1; }
  int x;
};

void hello_world(S s);
```

```c++
// hello_world.cpp

#include "hello_world.h"

#include <cstdio>

void hello_world2(S s) { printf("hello_world2: %d\n", s.x); }

void hello_world(S s) {
  printf("hello_world: %d\n", s.x);
  hello_world2(s);
}
```

```carbon
// main.carbon

library "Main";

import Cpp library "hello_world.h";

fn Run() -> i32 {
  var s : Cpp.S;
  Cpp.hello_world(s);
  return 0;
}
```

```shell
$ clang -c hello_world.cpp
$ bazel-bin/toolchain/carbon compile main.carbon
$ bazel-bin/toolchain/carbon link hello_world.o main.o --output=demo
$ ./demo
hello_world: 1
hello_world2: 1
```

Before this change (no thunk - copy constructor not called when calling
`hello_world()`):
```shell
$ ./demo
hello_world: -1219172304
hello_world2: 1
```
2025-08-06 10:27:33 +00:00