Decrease the size of nominal type descriptors and make them true-const by relative-addressing the other metadata they need to reference, which should all be included in the same image as the descriptor itself. Relative-referencing string constants exposes a bug in the Apple linker, which crashes when resolving relative relocations to coalesceable symbols (rdar://problem/22674524); work around this for now by revoking the `unnamed_addr`-ness of string constants that we take relative references to. (I haven't tested whether GNU ld or gold also have this problem on Linux; it may be possible to conditionalize the workaround to only apply to Darwin targets for now.)
A single extra inhabitant is good enough for the most important case,
that being a single level of optionality. Otherwise, we want to
reserve maximal flexibility for the implementation.
This commit also fixes a bug where I was not correctly defining
the extra-inhabitant rules for all of the existential cases.
Fixes <rdar://23122310> Runtime dynamic casts...
This makes runtime dynamic casts consistent with language rules, and
consequently makes specialization of generic code consistent with an
equivalent nongeneric implementation.
The runtime now supports casts from Optional<T> to U. Naturally the
cast fails on nil source, but otherwise succeeds if T is convertible to
U.
When casting T to Optional<U> the runtime succeeds whenever T is
convertible to U and simply wraps the result in an Optional.
To greatly simplify the runtime, I am assuming that
target-type-specific runtime cast entry points
(e.g. swift_dynamicCastClass) are never invoked with an optional
source. This assumption is valid for the following reasons. At the
language level optionals must be unwrapped before downcasting (via
as[?!]), so we only need to worry about SIL and IR lowering. This
implementation assumes (with asserts) that:
- SIL promotion from an address cast to a value casts should only happen
when the source is nonoptional. Handling optional unwrapping in SIL
would be too complicated because we need to check for Optional's own
conformances. (I added a test case to ensure this promotion does not
happen). This is not an issue for unchecked_ref_cast, which
implicitly unwraps optionals, so we can promote those!
- IRGen lowers unchecked_ref_cast (Builtin.castReference) directly to
a bitcast (will be caught by asserts).
- IRGen continues to emit the generic dynamicCast entry point for
address-casts (will be caught by asserts).
There was previously no way to detect a type that is nominally
Optional at runtime. The standard library, namely OutputStream, needs
to handle Optionals specially in order to cirumvent conversion to the
Optional's wrapped type. This should be done with conditional
conformance, but until that feature is available, Builtin.isOptional
will serve as a useful crutch.
Reuses the enum metadata layout and builder because most of the logic is
also required for Optional (generic arg and payload). We may want to
optimize this at some point (Optional doesn't have a Parent), but I
don't see much opportunity.
Note that with this approach there will be no change in metadata layout.
Changing the kind still breaks the ABI of course.
Also leaves the MirrorData summary string as "(Enum Value)". We should
consider changing it.
Fix spurious docs warning that @in and @in_guaranteed should be ``fn``.
Add ``#ifdef SWIFT_OBJC_INTEROP`` to silence a -Wunused-function
on linux since that function is only used from within that #ifdef
elsewhere.
Fix three -Wunused-function warnings on linux.
Fix two -Wunreachable-code warnings on linux dealing with
SWIFT_HAVE_WORKING_STD_REGEX.
This is a bit of a hodge-podge of related changes that I decided
weren't quite worth teasing apart:
First, rename the weak{Retain,Release} entrypoints to
unowned{Retain,Release} to better reflect their actual use
from generated code.
Second, standardize the names of the rest of the entrypoints around
unowned{operation}.
Third, standardize IRGen's internal naming scheme and API for
reference-counting so that (1) there are generic functions for
emitting operations using a given reference-counting style and
(2) all operations explicitly call out the kind and style of
reference counting.
Finally, implement a number of new entrypoints for unknown unowned
reference-counting. These entrypoints use a completely different
and incompatible scheme for working with ObjC references. The
primary difference is that the new scheme abandons the flawed idea
(which I take responsibility for) that we can simulate an unowned
reference count for ObjC references, and instead moves towards an
address-only scheme when the reference might store an ObjC reference.
(The current implementation is still trivially takable, but that is
not something we should be relying on.) These will be tested in a
follow-up commit. For now, we still rely on the bad assumption of
reference-countability.
The C++ ABI for static locals is a bit heavy compared to dispatch_once; doing this saves more than 1KB in runtime code size. Dispatch_once/call_once is also more likely to be hot because it's also used by Swift and ObjC code.
Alas, llvm::get_execution_seed() from llvm/ADT/Hashing.h still inflicts one static local initialization on us we can't override (without forking Hashing.h, anyway).
These functions are usually lowered away, but can remain if SIL optimizations run but not llvm optimizations.
This is only a workaround. I hope that I can remove the functions soon when we have a better support for buffers in SIL.
This reflects the fact that the attribute's only for compiler-internal use, and isn't really equivalent to C's asm attribute, since it doesn't change the calling convention to be C-compatible.
Getting the name of a type seems like reasonable core runtime functionality, and something the runtime can cache on its side too. Have the function return a pointer to a raw string in memory owned by the runtime, and have it be wrappen in a Swift.String on the standard library side.
We don't really need its peculiar behavior characteristics; its uses in the legacy mirror implementations can now be replaced by direct stringification of metatypes.
Move the ObjC internal declarations to a public runtime header so they can be shared, and rename _swift_deallocClassInstance to the more descriptive name swift_rootObjCDealloc (and make it only available with ObjC interop).
Set up a separate libSwiftStubs.a archive for C++ stub functionality that's needed by the standard library but not part of the core runtime interface. Seed it with the Stubs.cpp and LibcShims.cpp files, which consist only of stubs, though a few stubs are still strewn across the runtime code base.
The swift_unknown* entry points are not available on the Linux port.
Previously we would still attempt to use them in a couple of cases:
1) Foreign classes
2) Existentials and archetypes
3) Optionals of boxed existentials
Note that this patch changes IRGen to never emit the
swift_errorRelease/Retain entry points on Linux. We would like to
use them in the future if we ever adopt a tagged-pointer representation
for small errors. In this case, they can be brought back, and the
TypeInfo for optionals will need to be generalized to propagate the
reference counting of the payload type, instead of defaulting to
unknown if the payload type is not natively reference counted.
A similar change will need to be made to support blocks, if we ever
want to use the blocks runtime on Linux.
Fixes <rdar://problem/23335318>, <rdar://problem/23335537>,
<rdar://problem/23335453>.
This will let us eventually do tagged pointer optimization for small error values. We don't take advantage of this in IRGen yet, but we can take advantage of it in the dynamic cast code in a few places, so it gets exercised, and doing this now will let us backward-deploy the optimization when we do implement it in the future.
