These instructions do a bitcast operation without stack traffic (at the SIL level). unchecked_trivial_bit_cast represents a conversion from a potentially nontrivial type to a trivial type, such as from a class reference to Int. unchecked_ref_bit_cast represents a conversion between types for which retain_value and release_value has equivalent effects when applied on the input or output values.
Swift SVN r19053
These changes prevent a certain class of bogus errors, as well as several crashers. Unfortunately, though, they don't quite get us to the point where we can broadly use recursively defined protocol requirements, in the standard library. (To do so would require significant changes across the entire stack.)
Swift SVN r19019
Mandatory-inlined (aka transparent functions) are still treated as if they
had the location and scope of the call site. <rdar://problem/14845844>
Support inline scopes once we have an optimizing SIL-based inliner
Patch by Adrian Prantl.
Swift SVN r18835
LLVM's system_error.h has been changed to forward to the standard
version of the same. Update usage for the minor API changes that this
entails.
Based in part on a patch by Justin Bogner.
Swift SVN r18832
Add objc_metatype_to_object and objc_existential_metatype_to_object to convert metatypes to AnyObject, and objc_protocol to get a reference to an @objc protocol descriptor as a Protocol class instance.
Swift SVN r18824
put the result in a different place.
WIP: no IRGen support yet.
This will eventually be the required form when casting
to an address-only type; the existing instructions will
have only scalar outputs.
Swift SVN r18780
The deserializer holds a reference to the deserialized SILFunction, which
prevents Dead Function Elimination from erasing them.
We have a tradeoff on how often we should clean up the unused deserialized
SILFunctions. If we clean up at every optimization iteration, we may
end up deserializing the same SILFunction multiple times. For now, we clean
up only after we are done with the optimization iteration.
rdar://17046033
Swift SVN r18697
In Swift, a type and a value (such as a function) will never have the same
name, but we import C tag types (structs, unions, and enums) into the same
namespace as values, which can lead to ambiguity. If the user has managed
to make clear which one they want in source code, though, we should preserve
that information during serialization.
<rdar://problem/16979415>
Swift SVN r18536
the current SIL module, unless the AST sees them first.
This turns out to be important when we deserialize generic
reabstraction thunks, which have shared linkage so get serialized
again in the current SIL module.
There are some fundamental inconsistencies in the way parameter type
decls are handled, but we need a quick workaround to unblock progress
on the stdlib.
Fixes
<rdar://problem/16807985> Building Foundation overlay crashes in stdlib deserialization in r17377
Swift SVN r18173
This performs very conservative dependency generation for each compile task
within a full compilation. Any source file, swiftmodule, or Objective-C
header file that is /touched/ gets added to the dependencies list, which
is written out on a per-input basis at the end of compilation.
This does /not/ handle dependencies for the aggregated swiftmodule, swiftdoc,
generated header, or linked binary. This is just the minimum needed to get
Xcode to recognize what needs to be rebuilt when a header or Swift source
file changes. We can revisit this later.
This finishes <rdar://problem/14899639> for now.
Swift SVN r18045
This mostly falls out from the metatype cast infrastructure, but we need to generalize some Sema and SILGen code to accept AnyMetatypeType. Concrete-to-existential metatypes will need more runtime checking that isn't implemented, so raise a 'not implemented' error on those for now.
Swift SVN r17798
This hardcodes the list of platforms into unique identifiers, which
has tradeoffs to using strings. The reality is that various
logic in the compiler needs to be present to identify a given
platform, so hardcoding this list for now is ok. We can evaluate
in the future.
This change is motivated by some needed cleanup, supporting
App Extensions @availability, and unifying build configurations
with the platform names for @availability.
As part of this change, it introduces some slightly better
parser recovery for bad @availability attributes.
Moreover, the 'macosx' platform is renamed to 'osx'. We will need
to handle this mapping in the Clang Importer, but this is the
right thing to do.
Swift SVN r17767
When a module built with -autolink-force-load is imported, add a reference
to a special symbol in the corresponding library so that ld is forced to
link it.
This means the library will be linked into the final binary even if no other
symbols are used (which happens for some of our overlays that just add
category methods to Objective-C classes).
Second part of <rdar://problem/16829587>
Swift SVN r17751
We used to just use FileContext at deserialization time. That causes
multiple NormalProtocolConformances with the same protocol and type,
but from different modules. And we will have issues at IRGen.
With this commit, we serialize the parent module and deserialize the parent
module, and we are not going to have multiple NormalProtocolConformances with
the same protocol and type.
rdar://16809311
Swift SVN r17675
This allows a loaded swiftmodule to refer to decls that come from a header
imported by the module itself. There's an unfortunate sharing of the
namespace here with any /other/ headers loaded by /other/ modules (including
the module currently being compiled), but that shouldn't be too much of an
issue in practice -- the declarations would need to be compatible anyway if
you were doing the same thing in Objective-C.
More of <rdar://problem/16702101>. This is possibly the end of the compiler
work, but I'd like to add more tests to make sure everything is working.
Swift SVN r17639
This doesn't handle cross-references to decls /loaded/ from the header
just yet, so all that's testable right now is whether the header's imports
are visible from the secondary target (after being imported in response
to loading the serialized module).
More of <rdar://problem/16702101>
Swift SVN r17638
All serialization should go through serialize(). We don't currently support
serializing docs without serializing a module.
Also, tidy up how Serializer is used within Serialization.cpp.
Swift SVN r17637
It's not clear if all of these are always initialized before their
uses, so it seems reasonable to default initialize them to
predictable behavior.
Swift SVN r17612
It is reasonable that one module has a declaration for a SILFunction while
another module has the definition. We return from lookupSILFunction when
we find a definition and keep looking when a declaration is found from
a module.
rdar://16815627
Swift SVN r17586
We were accidentally forcing all members of a class to be instantiated in two places:
- by trying to look up an existing destructor decl in the class, and
- by adding the implicit destructor to the class, because addMember needlessly called loadAllMembers.
Fix the former problem by adding a 'has destructor' bit to ClassDecl so we can track whether the implicit destructor needs to be added without querying its members. Fix the latter by making IterableDeclContext::addMember not call loadAllMembers, and making loadAllMembers not barf when it sees existing members in the context.
Together with Jordan and JoeP's changes, this makes many interpreter tests now compile 3-20x faster.
Swift SVN r17562
This basically just means "it's a CF class" for now,
but you could imagine applying this to all sorts of
class-like types from peer runtimes that we can't
support all possible language features for.
There are quite a few language features that require
fairly deep object-model integration to implement,
like subclassing and adding polymorphic methods.
Some of those features, like final classes, are useful
to generally support as attributes, but most of
them aren't. At least in the short term, it makes
sense to have a big hammer we can hit things with.
Swift SVN r17428
Subscript declarations were still encoding the names of index
variables in the subscript type, which unintentionally made them
keyword arguments. Bring subscript declarations into the modern day,
using compound names to encode the subscript argument names, which
provides consistency for the keyword-argument world
<rdar://problem/14462349>. Note that arguments in subscripts default
to not being keyword arguments, which seems like the right default.
We now get keyword arguments for subscripts, so one can overload
subscripts on the names of the indices, and distinguish at the call
site. Under -strict-keyword-arguments, we require strictness here as well.
The IRGen/IDE/SILGen test updates are because the mangling of common
subscripts changed from accidentally having keyword arguments to not
having keyword arguments.
Swift SVN r17393
This is needed for unit tests to import app modules, but could also be used
to get rid of the "/32" used for the standard libraries.
This is the rest of <rdar://problem/16776466>
Swift SVN r17177
When deserializing, the serialized sil loader attempts to first lookup
from its SILModule the witness table to attempt to grab a witness table
declaration. Before this patch if the SILModule could not find it, it
would attempt to deserialize it. In certain edge cases this would cause
us to attempt to deserialize a function definition which violates the
invariant that that should never happen.
This patch adds an argument to SILModule::lookUpWitnessTable that
enables you to turn off the lazy deserialization behavior. The default
argument gives the current behavior which should be used everywhere
except for the deserializer.
Swift SVN r16740
An unsafe cast from a base to a derived class isn't really all that different from one from Builtin.NativeObject to an arbitrary class, so relax this pair of instructions to allow an arbitrary bitcast. This only combines the instructions; it doesn't attempt to simplify any codegen that was emitting round-trip casts before yet.
Swift SVN r16736
