Add global accessors to symbol list if VarDecl is fragile, i.e.
is non-resilient or its defining module allows non-resilient
access.
Don't set the class decl to hidden if it's in a package resilience
domain; even though its defining module is built resilently, the
class symbol should be visible across modules if they are in the
same package with resilience-bypass optimization. In such case,
treat its SubclassScope to Internal.
Resolves rdar://127321129
Quoting Slava:
"The AST-level access is allowed to be 'more public'. I honestly don't
know why, but it's always worked this way and the 'lowered' access
levels must always intersect the access levels of the parents but with
@uic, @aeic and @inlinable, that means just ignoring those attributes if
some enclosing context is not @uic"
rdar://128270848
Some notes:
1. If the result is non-Sendable and we didn't infer something that is
transferring, we still emit the current sema error that says that one cannot
assign a non-Sendable value to an async let.
2. When region isolation is enabled, but transferring args and results are
disabled, we leave the async let semantics alone. This means that the async let
closure is still @Sendable and one cannot pass in non-Sendable values to it.
Decls with a package access level are currently set to public SIL
linkages. This limits the ability to have more fine-grained control
and optimize around resilience and serialization.
This PR introduces a separate SIL linkage and FormalLinkage for
package decls, pipes them down to IRGen, and updates linkage checks
at call sites to include package linkage.
Resolves rdar://121409846
This allows calling a C++ function with default arguments from Swift without having to explicitly specify the values of all arguments.
rdar://103975014
When @_objcImplementation is used to implement a C function marked with __asm__, take the attribute into account when mangling SIL references.
In theory this change should also make `clang::OverloadableAttr` functions mangle correctly, but in practice the matching logic for @_cdecl @_objcImplementation doesn’t currently support overloadable functions (and it’s not wise to try anyway, since clang doesn’t promise that their ABI won’t change).
Fixes rdar://120503717.
https://github.com/apple/swift/pull/70100 prohibited `package` declarations from
ever being serialized in order to solve a problem in which the declarations
were being serialized inappropriately. That's too heavy handed, though, because
an `@_alwaysEmitIntoClient` function with `package` access *must* be serialized
because it has public non-abi linkage.
Resolves rdar://104711625
This accidentally started happening when I adjusted getEffectiveAccess to return `Public` for `Package` declarations in #69709. As a result, the optimizer thought it had more opportunities to go after declarations that used to be opaque. Unfortunately, this resulted in a miscompile as the devirtualizer was able to look through now-serialized package (static) functions. In one specific instance, the optimizer created direct calls to hidden accessors instead of going through the dispatch thunk.
Reformatting everything now that we have `llvm` namespaces. I've
separated this from the main commit to help manage merge-conflicts and
for making it a bit easier to read the mega-patch.
This is phase-1 of switching from llvm::Optional to std::optional in the
next rebranch. llvm::Optional was removed from upstream LLVM, so we need
to migrate off rather soon. On Darwin, std::optional, and llvm::Optional
have the same layout, so we don't need to be as concerned about ABI
beyond the name mangling. `llvm::Optional` is only returned from one
function in
```
getStandardTypeSubst(StringRef TypeName,
bool allowConcurrencyManglings);
```
It's the return value, so it should not impact the mangling of the
function, and the layout is the same as `std::optional`, so it should be
mostly okay. This function doesn't appear to have users, and the ABI was
already broken 2 years ago for concurrency and no one seemed to notice
so this should be "okay".
I'm doing the migration incrementally so that folks working on main can
cherry-pick back to the release/5.9 branch. Once 5.9 is done and locked
away, then we can go through and finish the replacement. Since `None`
and `Optional` show up in contexts where they are not `llvm::None` and
`llvm::Optional`, I'm preparing the work now by going through and
removing the namespace unwrapping and making the `llvm` namespace
explicit. This should make it fairly mechanical to go through and
replace llvm::Optional with std::optional, and llvm::None with
std::nullopt. It's also a change that can be brought onto the
release/5.9 with minimal impact. This should be an NFC change.
Previously, typechecking and SILGen would treat a function body as fragile as long as the declaration had a `@backDeployed` attribute, regardless of the platform specified by the attribute. This was overly conservative since back deployed functions are only emitted into the client on specific platforms. Now a `@backDeployed` function can reference non-`public` declarations on the platforms it is resilient on:
```
@backDeployed(before: iOS 15)
public func foo() {
#if os(iOS)
// Fragile; this code may be emitted into the client.
#else
// Resilient; this code won't ever be exposed to clients.
#endif
}
```
Resolves rdar://105298520
A new `RuntimeAttributeGenerator` is used to reference runtime
attribute generator functions synthesized by SILGen.
`#function` magic literal points to the declaration that declaration
attribute is attached to.
This replaces some of the existing checks, and
ensures that we don't emit coverage maps for
property wrapper backing initializers that don't
contain user code.
rdar://99931619
Previously we would delay the emission of
lazy variable getters and stored property
initializers for property wrapper backing storage.
This could lead to their definitions being dropped
if unused, meaning that we wouldn't run the
mandatory diagnostics passes over them.
Fix the logic such that we consider such cases as
having user-written code, and account for a couple
of cases where we can delay emission where we
didn't previously. There are more cases we can
handle here, but I'm leaving that as future work
for now, as `emitOrDelayFunction` is currently
only used for a handful of SILDeclRef kinds.
This is a source breaking change, but only for
invalid (albeit unused) code.
rdar://99962285
When computing the availability of a `SILFunction` for linkage the OS version specified in the `@_backDeploy` attribute should be preferred over the version in the `@available` attribute. This ensures that the corresponding symbol is weakly linked when deploying to older OSes than the back deploy "before" version.
Resolves rdar://99962885
A public designated initializer of a class would have its allocating
entry-point serialized in the module, meaning with `-O` that entry-point
can get inlined into programs linking against that module. Once that
entry-point is inlined, the program will _require_ that it remain non-delegating,
because it will depend on the 2nd entry-point (for actual initializing) to be in the
library.
As a result of this change, public initializers of an actor should be resilient in a
library, whether their underlying implementation is delegating or not.
When a distributed-actor-isolated witness is provided for a
non-distributed-actor-isolated requirement, always hop to the
distributed thunk.
While here, mark distributed thunks as being `final`, to ensure that
we always statically call them vs. trying to retrieve them from the
vtable (since they aren't in the vtable).
Previously we gave them the same SIL linkage as the method, then changed
the LLVM IR linkage to 'internal' (which is roughly equivalent to
SIL 'private') in IRGen.
This would crash in the SIL verifier if an @objc method was
'@_alwaysEmitIntoClient'. While such a combination of attributes is
silly since '@objc' methods are intrinsically part of the ABI, we
should not crash in this case.
The simplest fix is to just set the linkage to private at the SIL
level, avoiding the IRGen hack entirely.
