The main point of this change is to make sure that a shared function always has a body: both, in the optimizer pipeline and in the swiftmodule file.
This is important because the compiler always needs to emit code for a shared function. Shared functions cannot be referenced from outside the module.
In several corner cases we missed to maintain this invariant which resulted in unresolved-symbol linker errors.
As side-effect of this change we can drop the shared_external SIL linkage and the IsSerializable flag, which simplifies the serialization and linkage concept.
The SILGen testsuite consists of valid Swift code covering most language
features. We use these tests to verify that no unknown nodes are in the
file's libSyntax tree. That way we will (hopefully) catch any future
changes or additions to the language which are not implemented in
libSyntax.
...and "So" for the Swift-side representation of an Objective-C-side
conformance. These manglings are never used in public symbols, but
they're stil good to be consistent about.
This makes them consistent no matter what shenanigans are pulled by
the importer, particularly NS_ENUM vs. NS_OPTIONS and NS_SWIFT_NAME.
The 'NSErrorDomain' API note /nearly/ works with this, but the
synthesized error struct is still mangled as a Swift declaration,
which means it's not rename-stable. See follow-up commits.
The main place where this still falls down is NS_STRING_ENUM: when
this is applied, a typedef is imported as a unique struct, but without
it it's just a typealias for the underlying type. There's also still a
problem with synthesized conformances, which have a module mangled
into the witness table symbol even though that symbol is linkonce_odr.
rdar://problem/31616162
The witness table had shared linkage, but we weren't serializing them,
which would cause linking errors if we emitted a reference to such a
witness table from a different module than the one where it was first
defined, as a result of deserializing and optimizing SIL.
This issue was introduced when SIL witness table serialization was
made conditional on the -sil-serialize-witness-tables flag, which is
normally only enabled for the standard library.
When the flag was added, existing tests were updated to pass the
flag, which masked the issue. Remove the flag from existing tests,
ensure that imported witness tables are still [serialized], and add
a new test specifically for the behavior enabled by this flag.
This reverts commit 25985cb764. For now,
we're trying to avoid spurious non-structural changes to the mangling,
so that the /old/ mangling doesn't appear to change. That doesn't mean
no changes at all, but we can save this one for later.
This fixes a crash when referencing partially-applied methods
from @_inlineable functions.
Also, curry thunks for private methods do not need shared
linkage; private is sufficient.
Also, add a third [serializable] state for functions whose bodies we
*can* serialize, but only do so if they're referenced from another
serialized function.
This will be used for bodies synthesized for imported definitions,
such as init(rawValue:), etc, and various thunks, but for now this
change is NFC.
Also, mark witness thunks for [fragile] witnesses as [fragile].
This allows us to serialize the witness table, as well as any thunks
for witnesses declared @_transparent and @inline(__always).
Sema models enum case constructors as ApplyExprs. Formerly SILGen
would emit a case constructor function for each enum case,
constructing the enum value in the constructor body. ApplyExprs
of case constructors were lowered like any other call.
This is nice and straightforward but has several downsides:
1) Case constructor functions are very repetitive and trivial,
in particular for no-payload cases. They were declared
@_transparent and so were inlined at call sites, but for
public enums they still had to be emitted into the final
object file.
2) If the enum is generic, the substituted type may be loadable
even if the unsubstituted type is not, but since the case
constructor is polymorphic we had to allocate stack buffers
anyway, to pass the payload and result at the right abstration
level. This meant that for example Optional.Some(foo)
generated less-efficient SIL than the equivalent implicit
conversion.
3) We were missing out on peephole optimization opportunities when
the payload of an indirect case or address-only enum could be
emitted directly into the destination buffer, avoiding a copy.
One example would be when an enum payload is the result of
calling a function that returns an address-only value indirectly.
It appears we had unnecessary copies and takes even with -O.
Again, optional implicit conversions special-cased this.
This patch implements a new approach where a fully-formed call to
a element constructor is handled via a special code path where
the 'enum' or 'init_enum_data_addr' / 'inject_enum_addr'
instructions are emitted directly. These always work on the
substituted type, avoiding stack allocations unless needed.
An additional optimization is that the ArgumentSource abstraction
is used to delay evaluation of the payload argument until the
indirect box or address-only payload was set up.
If a element constructor is partially applied, we still emit a
reference to the constant as before.
It may seem like case constructor functions are at least useful
for resilience, but case constructors are transparent, so making
them resilient would require a new "transparent but only in this
module, and don't serialize the SIL body" declaration.
@inline(always) is almost what we need here, but this affect
mandatory inlining, only the optimizer, so it would be a
regression for non-resilient enums, or usages of resilient enums
in the current module.
A better approach is to construct resilient enums with a new
destructiveInjectEnumTag value witness function, which is
coming soon, and the general improvement from that approach
is what prompted this patch.
And include some supplementary mangling changes:
- Give the first generic param (depth=0, index=0) a single character mangling. Even after removing the self type from method declaration types, 'Self' still shows up very frequently in protocol requirement signatures.
- Fix the mangling of generic parameter counts to elide the count when there's only one parameter at the starting depth of the mangling.
Together these carve another 154KB out of a debug standard library. There's some awkwardness in demangled strings that I'll clean up in subsequent commits; since decl types now only mangle the number of generic params at their own depth, it's context-dependent what depths those represent, which we get wrong now. Currying markers are also wrong, but since free function currying is going away, we can mangle the partial application thunks in different ways.
Swift SVN r32896
Canonical dependent member types are always based from a generic parameter, so we can use a more optimal mangling that assumes this. We can also introduce substitutions for AssociatedTypeDecls, and when a generic parameter in a signature is constrained by a single protocol, we can leave that protocol qualification out of the unsubstituted associated type mangling. These optimizations together shrink the standard library by 117KB, and bring the length of the longest Swift symbol in the stdlib down from 578 to 334 characters, shorter than the longest C++ symbol in the stdlib.
Swift SVN r32786
'Ss' appears in manglings tens of thousands of times in the standard library and is also incredibly frequent in other modules. This alone is enough to shrink the standard library by 59KB.
Swift SVN r32409
The old code predates NumberLiteralExpr having a "negative" field.
Fixing this avoids creating a temporary signed integer with a value of
INT_MAX+1 when trying to compute INT_MIN.
rdar://problem/21680700
Swift SVN r30814
If a SILDeclRef references a decl that isn't explicit in the source code and can't be referenced externally, then we only need to emit it if it's referenced in the current TU. Partially addresses rdar://problem/21444126, though we still eagerly generate witness tables for generated conformances, which still pull in a bunch of noise.
Swift SVN r29536
The rule changes are as follows:
* All functions (introduced with the 'func' keyword) have argument
labels for arguments beyond the first, by default. Methods are no
longer special in this regard.
* The presence of a default argument no longer implies an argument
label.
The actual changes to the parser and printer are fairly simple; the
rest of the noise is updating the standard library, overlays, tests,
etc.
With the standard library, this change is intended to be API neutral:
I've added/removed #'s and _'s as appropriate to keep the user
interface the same. If we want to separately consider using argument
labels for more free functions now that the defaults in the language
have shifted, we can tackle that separately.
Fixes rdar://problem/17218256.
Swift SVN r27704
Import our fake Foundation rather than redeclaring NSString in SILGen's
fake "gizmo" module...then fix up other tests to more carefully use SILGen's
mock SDK in avoiding the real set of overlays...using -enable-source-import.
In the mid-term SILGen should probably switch to using the shared
clang-importer-sdk in test/Inputs/. In the long term we need that to be
using proper modules too rather than -enable-source-import.
No intended functionality change, but without this the next commit breaks
things (because members are deserialized less eagerly).
Swift SVN r27195
This lets us disambiguate the symbols for static and instance properties, and enables us to eventually leave the useless "self" type mangling out of method symbols. Fixes rdar://19012022 and dupes thereof, including crasher #1341.
Swift SVN r25111
This will have an effect on inlining into thunks.
Currently this flag is set for witness thunks and thunks from function signature optimization.
No change in code generation, yet.
Swift SVN r24998
Most tests were using %swift or similar substitutions, which did not
include the target triple and SDK. The driver was defaulting to the
host OS. Thus, we could not run the tests when the standard library was
not built for OS X.
Swift SVN r24504
Doing so is safe even though we have mock SDK. The include paths for
modules with the same name in the real and mock SDKs are different, and
the module files will be distinct (because they will have a different
hash).
This reduces test runtime on OS X by 30% and brings it under a minute on
a 16-core machine.
This also uncovered some problems with some tests -- even when run for
iOS configurations, some tests would still run with macosx triple. I
fixed the tests where I noticed this issue.
rdar://problem/19125022
Swift SVN r23683
Per API review with Ali. While we're here, give the initializer a corresponding 'rawValue' argument label, and change the associated type name to RawValue to match.
Swift SVN r21888
Redefine the RawRepresentable protocol to use an 'init?' method instead of 'fromRaw(Raw)', and a 'raw' get-only property instead of 'toRaw()'. Update the compiler to support deriving conformances for enums and option sets with the new protocol. rdar://problem/18216832
Swift SVN r21762
