Rename the funnel points for demangling strings/nodes to metadata to
swift_getTypeByMangled(Name|Node) and make them overridable. This will let
us back-deploy mangling improvements and bug fixes.
Simplify the signature of the internal _getTypeByMangledName() used by the
standard library to what we actually (currently) use. Drop it as a
compatibility override, because it’s not a useful place to introduce
customization.
Clean up the interfaces used to go from a mangled name or demangle tree to
metadata. Parameterize these interfaces for generic parameter substitutions
(already in use) and dependent conformance substitutions (speculative).
When we emit "false" symbolic references to accesors for conformances or
type metadata, ensure that we end up with two-byte-aligned symbolic
references. This addresses a problem with ARM+Thumb compilation where the
low bit wasn't getting set for Thumb code.
Fixes rdar://problem/46067353.
The patched +[NSBundle bundleForClass:] crashes on Nil, while the original Foundation implementation returns the main bundle. Avoid the crash and pass Nil through to Foundation. This also ensures that Nil is passed through to class_getImageName rather than crashing.
SR-9188 rdar://problem/45849924
Split these two functions into a fast path (for the cached case) and a slow
path. Make the slow path overridable, so we can patch it in the future if
needed.
When we are looking for the specific type for a protocol conformance (e.g.,
because we may have a subclass of the type that declared conformances), don't
go back through swift_conformsToProtocol() multiple times, which
requires more lookups in the global conformance table. Instead, use
the (known) protocol conformance descriptor.
Previously, they would forward their unused spare bits to be used by other multi-payload enums, but
did not implement anything for single-payload extra inhabitants.
The current representation of an associated conformance in a witness
tables (e.g., Iterator: IteratorProtocol within a witness table for
Sequence) is a function that the client calls.
Replace this with something more like what we do for associated types:
an associated conformance is either a pointer to the witness table (once
it is known) or a pointer to a mangled name that describes that
conformance. On first access, demangle the mangled name and replace the
entry with the resulting witness table. This will give us a more compact
representation of associated conformances, as well as always caching
them.
For now, the mangled name is a sham: it’s a mangled relative reference to
the existing witness table accessors, not a true mangled name. In time,
we’ll extend the support here to handle proper mangled names.
Part of rdar://problem/38038799.
Previously, the stdlib provided:
- getters for AnyKeyPath and PartialKeyPath, which have remained;
- a getter for KeyPath, which still exists alongside a new read
coroutine; and
- a pair of owned mutable addressors that provided modify-like behavior
for WritableKeyPath and ReferenceWritableKeyPath, which have been
replaced with modify coroutines and augmented with dedicated setters.
SILGen then uses the most efficient accessor available for the access
it's been asked to do: for example, if it's been asked to produce a
borrowed r-value, it uses the read accessor.
Providing a broad spectrum of accessor functions here seems acceptable
because the code-size hit is fixed-size: we don't need to generate
extra code per storage declaration to support more alternatives for
key paths.
Note that this is just the compiler ABI; the implementation is still
basically what it was. That means the implementation of the setters
and the read accessor is pretty far from optimal. But we can improve
the implementation later; we can't improve the ABI.
The coroutine accessors have to be implemented in C++ and used via
hand-rolled declarations in SILGen because it's not currently possible
to declare independent coroutine accessors in Swift.
Introduce a new runtime entry point, swift_getAssociatedConformanceWitness(),
which extracts an associated conformance witness from a witness table.
Teach IRGen to use this entry point rather than loading the witness
from the witness table and calling it directly.
There’s no advantage to doing this now, but it is staging for changing the
representation of associated conformances in witness tables.
TargetGenericParamRef is a specialized structure used to describe the
subject of a generic requirement, e.g., the “T.Assoc” in “T.Assoc: P”.
Replace it with a mangled name, for several reasons:
1) Mangled type names are also fairly concise, can often be shared, and
are a well-tested path
2) Mangled type names can express any type, which might be useful in the
future
3) This structure doesn’t accommodate specifically stating where the
conformances come from (to extract associated type witnesses). Neither
can mangled names, but we’d like to do that work in only one place.
This change exposed an existing bug where we improperly calculated the
generic parameter counts for extensions of nested generic types. Fix that
bug here (which broke an execution test).
The standard library never ended up needing the low extra inhabitants (<4G on 64-bit Darwin,
<4K elsewhere), so BridgeObject can have the same set of extra inhabitants as the other refcounted
types, allowing `String?????` and `Array??????????` to still use optimized representations.
rdar://problem/45881464
The “old” mangling that is used for runtime names of @objc protocols
uses a simpler substitution scheme, so you can’t simply take a mangled
name from the new mangling and fix up the ends.
Fixes rdar://problem/45685649.
Silence warnings about deleted defaulted constructors due to the
non-trivial constructor for the atomic type. Guard a conditionally used
function with the appropriate guard.
Right now we expect that every class and protocol has a field
descriptor that tells us if the entity is @objc or not.
For imported types, the descriptor will not exist if we did not
directly emit a field whose concrete type contains the imported
type. For example, in lldb, we might have a generic type whose
runtime substituted type includes an imported type.
In this case, TypeLowering would fail to produce a layout because
it did not find a field descriptor for the imported type.
A better approach is to have the TypeDecoder call a different
factory method for imported types, and handle them specially in
TypeLowering, bypassing the field type metadata altogether.
When checking conformance requirements against an @objc protocol, also
check for an @objc existential using protocol_conformsToProtocol().
Fixes rdar://problem/45685649.
Rather than use a function, use a templated structure with a constant
expression to metaprogram the alignment of the type. NFC.
Adjust the templating for the SIMD vector types. Use the reserved
spelling for the clang extension, and rename the types. NFC.
The meat of this change is the explicit template specialization for the
float 3 vector and double 3 vector. Since they are going to be emitted
as float 4 vector underneath everything, we can simply treat them the
same. This has a benefit of allowing us to share the same
specialization of the witness tables. Additionally, it works around a
bug in clang where we cannot correctly decorate the extended type with
Microsoft's ABI (which is a separate issue). Since this saves some
bytes, this is still beneficial.
The key thing here is that all of the underlying code is exactly the same. I
purposely did not debride anything. This is to ensure that I am not touching too
much and increasing the probability of weird errors from occurring. Thus the
exact same code should be executed... just the routing changed.
We were strangely excluding protocols from being symbolically referenced
in the any-generic-type production, which meant that we could not resolve
(e.g.) associated type references to private protocols at runtime. Allow
protocol symbolic references in this position, and cope with it in the
demangler.
Fixes the rest of rdar://problem/44977236.
Previously we were only doing this in assert builds, or if the
new Objective-C runtime metadata update hook mechanism was
available.
However, swift_checkMetadataState() gets called on the superclass
of a class, and it assumes that the metadata entry for the
superclass already exists in the singleton cache.
So even on an older runtime when there's no initialization work
to be done, we have to realize the superclass to populate the
singleton cache so that the check can succeed.
Fixes <rdar://problem/45569020>.
