Rather than swizzle the superclass of these bridging classes at +load time, have the compiler set their ObjC runtime base classes, using a "@_swift_native_objc_runtime_base" attribute that tells the compiler to use a different implicit base class from SwiftObject. This lets the runtime shed its last lingering +loads, and should overall be more robust, since it doesn't rely on static initialization order or deprecated ObjC runtime calls.
Swift SVN r28219
Configure the runtime to build with -Wglobal-constructors, and Lazy-fy almost everything that gets flagged. (I gave "swift_isaMask" a pass since that's almost definitely hot enough to warrant a static initialization.) Make some improvements to the Lazy wrapper, using aligned_storage to ensure that it's trivially constructed and destructed.
Swift SVN r28199
The __lldb_expr modules are special as in they are autogenerated by LLDB and meant to not be user-accessible, so showing them adds visual noise for no user benefit
I am open to the notion of adding a flag to swift-demangle to the same effect, but that seems much lower priority
Swift SVN r28195
This matches how dispatch_once works in C, dramatically cutting the cost of a global accessor by avoiding the runtime call in the hot path and giving the global a unique branch for the CPU to predict away. For now, only do this for Darwin; non-ObjC platforms don't necessarily expose their "done" value as ABI like ours do.
While we're here, change "once" to take a thin function pointer. We don't ever emit global initializers with context dependencies, and this simplifies the runtime glue between swift_once and dispatch_once/std::call_once a bit.
Swift SVN r28166
This has a couple benefits:
- Since metadata allocations are already guarded by a lock, the allocator doesn't require synchronization, and can be much much simpler and a little faster than malloc.
- By bypassing malloc, we also avoid tools like 'heap' prying into our metadata cache and misrepresenting cache entries keyed on classes as live objects, fixing rdar://problem/20562886.
In my unscientific local tests, this appeared to give a small across-the-board improvement to Onone performance in the perf test suite, though not far enough from noise for me to declare that definitively. Fixing the bug is the bigger point here.
Swift SVN r27856
and use it to update LeaksChecker in a robust way to handle new metadata
kinds.
I also fixed a small typo where the native ErrorType was not included in
the range of non-ObjC isa metadata kinds.
Swift SVN r27718
We have enough flag bits on function types now to warrant stashing an extra word in the metadata key alongside the arguments and results, so add one, and pack the number of arguments, function convention, and 'throws' bit in there. This lets us merge the separate metadata caches for thick/thin/block/C functions into one, saving a bit of runtime memory, and simplifying a bunch of repetitive code in the runtime and IRGen.
This also fixes a subtle bug we had where the runtime getFunctionTypeMetadata function expected the result argument to be passed in the arguments array, but IRGen was passing it as a separate argument, which would have caused function type metadata to fail to be uniqued by result type.
Swift SVN r27651
Fixes rdar://problem/20583365, and incidentally gets test/Interpreter/layout_reabstraction.swift to work without ObjC interop as well.
Swift SVN r27557
Following Dmitri's advice, remove an unneeded dispatch_once() and use an
initializer of a function-local static variable instead. NFC.
Swift SVN r27516
@objc protocols aren't supported with an ObjC runtime, but we still want values of AnyObject type to be word-sized. Handle this by turning the binary "needsWitnessTable" condition into a "dispatch strategy" enum, so we can recognize the condition "has no methods, so neither swift nor objc dispatch" as distinct from either swift or ObjC protocol representations. Assign this dispatch strategy when we lower AnyObject. Should be NFC for the ObjC-enabled build.
(It would also be beneficial for the ObjC-runtime-enabled version of Swift if AnyObject weren't an @objc protocol; that would mean we could give it a canonical protocol descriptor in the standard library, among other things. There are fairly deep assumptions in Sema that AnyObject is @objc, though, and it's not worth disturbing those assumptions right now.)
Reapplying with updates to the runtime unit tests.
Swift SVN r27341
@objc protocols aren't supported with an ObjC runtime, but we still want values of AnyObject type to be word-sized. Handle this by turning the binary "needsWitnessTable" condition into a "dispatch strategy" enum, so we can recognize the condition "has no methods, so neither swift nor objc dispatch" as distinct from either swift or ObjC protocol representations. Assign this dispatch strategy when we lower AnyObject. Should be NFC for the ObjC-enabled build.
(It would also be beneficial for the ObjC-runtime-enabled version of Swift if AnyObject weren't an @objc protocol; that would mean we could give it a canonical protocol descriptor in the standard library, among other things. There are fairly deep assumptions in Sema that AnyObject is @objc, though, and it's not worth disturbing those assumptions right now.)
Swift SVN r27338
If we see an ErrorType-conforming value type while casting to NSError, do the coercion in the runtime, so that e.g. 'error as Any as! NSError' works as well as 'error as NSError'.
Unfortunately, we can't do this for ErrorType-conforming classes without shattering SIL's brittle view of the world due to rdar://problem/20507075. Casts between class types are fairly deeply assumed not to change the identity of the reference.
Swift SVN r27225
Some future-proofing to let us change ErrorType's reference counting in the future, or to use various tagged pointer optimizations in its representation.
Swift SVN r27213
We may want to use optimized ErrorType representations that don't naturally "become" NSErrors, such as tagged-pointer representations of small error enums, or a tagged function pointer to a deferred error type constructor. Rename the runtime function to something a bit more descriptive of its real purpose, not its implementation.
Swift SVN r27209
On OS X 10.9 and iOS 7 the runtime support for #availability was reading the system plist
on every query. Now, use dispatch_once to cache the results of the expensive
plist lookup ourselves.
rdar://problem/18471069
Swift SVN r27199
Simplify the observation evaluation in the ErrorType race test by using evaluateObservationsAllEqual, and run the test for real. Apparently round-trip NSString and NSDictionary bridging doesn't produce a stable object, so use valueForKey as a hack to sidestep bridging so we can verify that the identities of the bridged domain and code aren't racy.
Swift SVN r27003
Atomically initialize and load the NSError bridging fields within an ErrorType box so that we do the right thing when two threads concurrently coerce the box to NSError.
Swift SVN r26996
If the NSError part of the box hasn't been initialized yet, fill it in with the domain and code of the contained value. This will allow us to efficiently turn ErrorType values into NSErrors, either for bridging or for coercion purposes.
Swift SVN r26958
Previously some parts of the compiler referred to them as "fields",
and most referred to them as "elements". Use the more generic 'elements'
nomenclature because that's what we refer to other things in the compiler
(e.g. the elements of a bracestmt).
At the same time, make the API better by providing "getElement" consistently
and using it, instead of getElements()[i].
NFC.
Swift SVN r26894
This allows ErrorTypes referencing Cocoa NSErrors to be pattern-matched against our bridged enums, at least dynamically. (The compiler still doesn't understand these bridging conversions yet.)
Swift SVN r26839
