Type erasure requires a circular construction by its very nature:
@_typeEraser(AnyProto)
protocol Proto { /**/ }
public struct AnyProto : Proto {}
If we eagerly resolve AnyProto, the chain of resolution steps that
deserialization must make goes a little something like this:
Lookup(Proto)
-> Deserialize(@_typeEraser(AnyProto))
-> Lookup(AnyProto)
-> DeserializeInheritedStuff(AnyProto)
-> Lookup(Proto)
This cycle could be broken if the order of incremental inputs was
such that we had already cached the lookup of Proto.
Resolve this cycle in any case by suspending the deserialization of the
type eraser until the point it's demanded by adding
ResolveTypeEraserTypeRequest.
rdar://61270195
As part of this, we have to change the type export rules to
prevent `@convention(c)` function types from being used in
exported interfaces if they aren't serializable. This is a
more conservative version of the original rule I had, which
was to import such function-pointer types as opaque pointers.
That rule would've completely prevented importing function-pointer
types defined in bridging headers and so simply doesn't work,
so we're left trying to catch the unsupportable cases
retroactively. This has the unfortunate consequence that we
can't necessarily serialize the internal state of the compiler,
but that was already true due to normal type uses of aggregate
types from bridging headers; if we can teach the compiler to
reliably serialize such types, we should be able to use the
same mechanisms for function types.
This PR doesn't flip the switch to use Clang function types
by default, so many of the clang-function-type-serialization
FIXMEs are still in place.
To guard against the kind of complex failures seen in rdar://58690068,
improve this regression test with a copy of a shrunken version of
CocoaError.Code.
In that radar, CocoaError.Code was defined in one overlay and extended
in another translation unit. When deserialization encountered this, it
failed to search the overlay's nested type tables and fell through to
re-enter lookupDirect.
SIL type lowering erases DynamicSelfType, so we generate
incorrect code when casting to DynamicSelfType. Fixing this
requires a fair amount of plumbing, but most of the
changes are mechanical.
Note that the textual SIL syntax for casts has changed
slightly; the target type is now a formal type without a '$',
not a SIL type.
Also, the unconditional_checked_cast_value and
checked_cast_value_br instructions now take the _source_
formal type as well, just like the *_addr forms they are
intended to replace.
After setting up the .swiftsourceinfo file, this patch starts to actually serialize
and de-serialize source locations for declaration. The binary format of .swiftsourceinfo
currently contains these three records:
BasicDeclLocs: a hash table mapping from a USR ID to a list of basic source locations. The USR id
could be retrieved from the following DeclUSRs record using an actual decl USR. The basic source locations
include a file ID and the results from Decl::getLoc(), ValueDecl::getNameLoc(), Decl::getStartLoc() and Decl::getEndLoc().
The file ID could be used to retrieve the actual file name from the following SourceFilePaths record.
Each location is encoded as a line:column pair.
DeclUSRS: a hash table mapping from USR to a USR ID used by location records.
SourceFilePaths: a hash table mapping from a file ID to actual file name.
BasicDeclLocs should be sufficient for most diagnostic cases. If additional source locations
are needed, we could always add new source location records without breaking the backward compatibility.
When de-serializing the source location from a module-imported decl, we calculate its USR, retrieve the USR ID
from the DeclUSRS record, and use the USR ID to look up the basic location list in the BasicDeclLocs record.
For more details about .swiftsourceinfo file: https://forums.swift.org/t/proposal-emitting-source-information-file-during-compilation
Fixes rdar://55560962 and https://bugs.swift.org/browse/SR-11495
This bug was caused because we'd fail to lookup C in B, because we
hadn't computed its type yet and therefore B is filtered from lookup.
Just remove the filter for interface type.
This removes it from the AST and largely replaces it with AnyObject
at the SIL and IRGen layers. Some notes:
- Reflection still uses the notion of "unknown object" to mean an
object with unknown refcounting. There's no real reason to make
this different from AnyObject (an existential containing a
single object with unknown refcounting), but this way nothing
changes for clients of Reflection, and it's consistent with how
native objects are represented.
- The value witness table and reflection descriptor for AnyObject
use the mangling "BO" instead of "yXl".
- The demangler and remangler continue to support "BO" because it's
still in use as a type encoding, even if it's not an AST-level
Type anymore.
- Type-based alias analysis for Builtin.UnknownObject was incorrect,
so it's a good thing we weren't using it.
- Same with enum layout. (This one assumed UnknownObject never
referred to an Objective-C tagged pointer. That certainly wasn't how
we were using it!)
This flag is set by DefinitInitialization if the lifetime of the stored value is controlled dynamically.
If the flag is set, it's not (easily) possibly to statically calculate the lifetime of the stored value.
When the outermost property wrapper associated with a property has a
`wrapperValue`, create the projection property (with the `$` prefix)
at the same access level as the original property. This puts the
wrapped-value interface and the projection interface at the same level.
The newly-introduced @_projectionValueProperty attribute is implicitly
created to establish the link between the original property and the
projection value within module interfaces, where both properties will
be explicitly written out.
We don't need to serialize the protocol's superclass, we can compute it from the
generic signature. Previously, we would drop the superclass while
serializing because we didn't check the generic signature in
SuperclassTypeRequest, which would cause us to cache `NULL` when we
called `setSuperclass` for a protocol with a superclass constraint.
Fixes rdar://50526401
The merge-modules phase doesn't have an active type checker, so we were bailing
out of property-wrapper queries before we had a chance to check the cache.
Move the check later, to the points where we actually need a type checker.
Fixes SR-10844 / rdar://problem/51484958.
To distinguish between classes which have the same name (but are in different contexts).
Fixes a miscompile if classes with the same name are used from a different module.
SR-10634
rdar://problem/50538534
Serialize the relationship between a property that has an attached delegate
and its backing variable, so deserialization can reestablish that link.
Fixes rdar://problem/50447022.
Fix a trio of issues involving mangling for opaque result types:
* Symbolic references to opaque type descriptors are not substitutions
* Mangle protocol extension contexts correctly
* Mangle generic arguments for opaque result types of generic functions
The (de-)serialization of generic parameter lists for opaque type
declarations is important for the last bullet, to ensure that the
mangling of generic arguments of opaque result types works across
module boundaries.
Fixes the rest of rdar://problem/50038754.
This is an attribute that gets put on an import in library FooKit to
keep it from being a requirement to import FooKit. It's not checked at
all, meaning that in this form it is up to the author of FooKit to
make sure nothing in its API or ABI depends on the implementation-only
dependency. There's also no debugging support here (debugging FooKit
/should/ import the implementation-only dependency if it's present).
The goal is to get to a point where it /can/ be checked, i.e. FooKit
developers are prevented from writing code that would rely on FooKit's
implementation-only dependency being present when compiling clients of
FooKit. But right now it's not.
rdar://problem/48985979
If a value decl is internal hasTestableOrPrivateImport will succeed (or
fail) without looking at the filename. However this breaks when we query
an internal storage decl with private formal access for a private
setter: the query would fail because no filename was serialized for the
decl (we only serialize filenames for private decls). So in the special
case of a internal storage with private accessor also serialize the
filename.
rdar://48516545
I noticed that I enabled ownership verification on all of the simple run swift
tests, but I didn't on the simple build swift tests. I have prepared a commit
that enables that. This commit contains some test fixes needed to make it pass.
In a previous commit, I banned in the verifier any SILValue from producing
ValueOwnershipKind::Any in preparation for this.
This change arises out of discussions in between John, Andy, and I around
ValueOwnershipKind::Trivial. The specific realization was that this ownership
kind was an unnecessary conflation of the a type system idea (triviality) with
an ownership idea (@any, an ownership kind that is compatible with any other
ownership kind at value merge points and can only create). This caused the
ownership model to have to contort to handle the non-payloaded or trivial cases
of non-trivial enums. This is unnecessary if we just eliminate the any case and
in the verifier separately verify that trivial => @any (notice that we do not
verify that @any => trivial).
NOTE: This is technically an NFC intended change since I am just replacing
Trivial with Any. That is why if you look at the tests you will see that I
actually did not need to update anything except removing some @trivial ownership
since @any ownership is represented without writing @any in the parsed sil.
rdar://46294760
Otherwise, we can't represent a cross-reference to generic parameters
in a parent type /when used in an extension/.
https://bugs.swift.org/browse/SR-9084
Previously, the fast path for nested types only worked when the nested
type was defined in a Swift module or a Clang module without an
overlay; this is because it was originally designed to fix circularity
issues when merging partial modules for a single target. By having a
Swift overlay module pass through requests for nested types to the
underlying Clang module, we get the fast-path behavior in more cases.
(The one case where it /won't/ kick in is if the overlay has a nested
type that shadows a nested type from the Clang module, but that's
probably pretty rare!)
