The behaviour of ilist has changed in LLVM. It is no longer permissible to
dereference the `end()` value. Add a check to ensure that we do not
accidentally dereference the iterator.
This lets us get to the goal of +0 guaranteed closure contexts. NFC yet, just add the under-the-hood ability for partial_apply instructions producing callee-guaranteed closures to be parsed, printed, and serialized.
This patch is rather large, since it was hard to make this change
incrementally, but most of the changes are mechanical.
Now that we have a lighter-weight data structure in the AST for mapping
interface types to archetypes and vice versa, use that in SIL instead of
a GenericParamList.
This means that when serializing a SILFunction body, we no longer need to
serialize references to archetypes from other modules.
Several methods used for forming substitutions can now be moved from
GenericParamList to GenericEnvironment.
Also, GenericParamList::cloneWithOuterParameters() and
GenericParamList::getEmpty() can now go away, since they were only used
when SILGen-ing witness thunks.
Finally, when printing generic parameters with identical names, the
SIL printer used to number them from highest depth to lowest, by
walking generic parameter lists starting with the innermost one.
Now, ambiguous generic parameters are numbered from lowest depth
to highest, by walking the generic signature, which means test
output in one of the SILGen tests has changed.
This is the first, and most trivial, usage of the new
GenericSignature::getSubstitutions() method.
Note that getForwardingSubstitutions() now takes a
GenericSignature, which is slightly awkward.
However, this is in line with our goal of 'hollowing out'
GenericParamList by removing knowledge of the finalized
generic requirements.
Also, there is now a new getForwardingSubstitutionMap()
function, which returns an interface type substitution
mapping. This is used in the new getForwardingSubstitutions()
implementation, and all also be used elsewhere later.
Finally, in the SILFunction we now cache the forwarding
substitutions, instead of re-computing them every time.
I doubt this makes a big difference in performance, but
it's a simple enhancement and every little bit helps.
Previously, if a generic type had a stored property with
a generic type and an initializer expression, we would
emit the expression directly in the body of each designated
initializer.
This is a problem if the designated initializer is defined
within an extension (even in the same source file), because
extensions have a different set of generic parameters and
archetypes.
Also, we've had bugs in the past where emitting an
expression multiple times didn't work properly. While these
might currently all be fixed, this is a tricky case to test
and it would be best to avoid it.
Fix both problems by emitting the initializer expression
inside its own function at the SIL level, and call the
initializer function from each designated initializer.
I'm using the existing 'variable initializer' mangling for this;
it doesn't seem to be used for anything else right now.
Currently, the default memberwise initializer does not use
this, because the machinery for emitting it is somewhat
duplicated and separate from the initializer expressions in
user-defined constructors. I'll clean this up in an upcoming
patch.
Fixes <https://bugs.swift.org/browse/SR-488>.
Till now there was no way in SIL to explicitly express a dependency of an instruction on any opened archetypes used by it. This was a cause of many errors and correctness issues. In many cases the code was moved around without taking into account these dependencies, which resulted in breaking the invariant that any uses of an opened archetype should be dominated by the definition of this archetype.
This patch does the following:
- Map opened archetypes to the instructions defining them, i.e. to open_existential instructions.
- Introduce a helper class SILOpenedArchetypesTracker for creating and maintaining such mappings.
- Introduce a helper class SILOpenedArchetypesState for providing a read-only API for looking up available opened archetypes.
- Each SIL instruction which uses an opened archetype as a type gets an additional opened archetype operand representing a dependency of the instruction on this archetype. These opened archetypes operands are an in-memory representation. They are not serialized. Instead, they are re-constructed when reading binary or textual SIL files.
- SILVerifier was extended to conduct more thorough checks related to the usage of opened archetypes.
If a closure captures the dynamic 'Self' type, but no value of type 'Self'
(for example, it is possible to have a weak capture of 'self'; if the weak
reference becomes nil, there's no way for the closure to get the dynamic
'Self' type from the value).
In this case, add a hidden argument of type $Self.Type, and pass in the
Self metatype.
Fixes <https://bugs.swift.org/browse/SR-1558> / <rdar://problem/22299905>.
Till now there was no way in SIL to explicitly express a dependency of an instruction on any opened archetypes used by it. This was a cause of many errors and correctness issues. In many cases the code was moved around without taking into account these dependencies, which resulted in breaking the invariant that any uses of an opened archetype should be dominated by the definition of this archetype.
This patch does the following:
- Map opened archetypes to the instructions defining them, i.e. to open_existential instructions.
- Introduce a helper class SILOpenedArchetypesTracker for creating and maintaining such mappings.
- Introduce a helper class SILOpenedArchetypesState for providing a read-only API for looking up available opened archetypes.
- Each SIL instruction which uses an opened archetype as a type gets an additional opened archetype operand representing a dependency of the instruction on this archetype. These opened archetypes operands are an in-memory representation. They are not serialized. Instead, they are re-constructed when reading binary or textual SIL files.
- SILVerifier was extended to conduct more thorough checks related to the usage of opened archetypes.
Till now there was no way in SIL to explicitly express a dependency of an instruction on any opened archetypes used by it. This was a cause of many errors and correctness issues. In many cases the code was moved around without taking into account these dependencies, which resulted in breaking the invariant that any uses of an opened archetype should be dominated by the definition of this archetype.
This patch does the following:
- Map opened archetypes to the instructions defining them, i.e. to open_existential instructions.
- Introduce a helper class SILOpenedArchetypesTracker for creating and maintaining such mappings.
- Introduce a helper class SILOpenedArchetypesState for providing a read-only API for looking up available opened archetypes.
- Each SIL instruction which uses an opened archetype as a type gets an additional opened archetype operand representing a dependency of the instruction on this archetype. These opened archetypes operands are an in-memory representation. They are not serialized. Instead, they are re-constructed when reading binary or textual SIL files.
- SILVerifier was extended to conduct more thorough checks related to the usage of opened archetypes.
There was a weird corner case with nested generic functions that
would fail in the SIL verifier with some nonsense about archetypes
out of context.
Fix this the "right" way, by re-working Sema function declaration
validation to assign generic signatures in a more principled way.
Previously, nested functions did not get an interface type unless
they themselves had generic parameters.
This was inconsistent with methods nested inside generic types,
which did get an interface type even if they themselves did not
have a generic parameter list.
There's some spill-over in SILGen from this change. Mostly it
makes things more consistent and fixes some corner cases.
We now have enough machinery in place to reference local generic
functions which have captures, to get a value of function type
that can be passed around.
Generic local functions still cannot be directly called from
function call expressions, since those go down a different
path in SILGenApply.cpp -- the next patch will add support for
this case.
The verifier now asserts that Throws, ThrowsLoc and isBodyThrowing()
match up.
Also, add /*Label=*/ comments where necessary to make the long argument
lists easier to read, and cleaned up some inconsistent naming conventions.
I caught a case where ClangImporter where we were passing in a loc as
StaticLoc instead of FuncLoc, but probably this didn't affect anything.
BoundGenericType::getSubsitutions() would only look at the bound
generic arguments of the innermost type, ignoring parent types.
However, it would then proceed to walk the AllArchetypes list
of all outer generic parameter lists when forming the final
result.
The gatherAllSubstitutions() would also walk through parent types.
As a result, outer generic parameters would appear multiple
times.
Simplify gatherAllSubstitutions() to just skip to the innermost
BoundGenericType, and delegate to getSubsitutions() for the rest.
Most calls to gatherAllSubstitutions() are from SILGen it seems,
and fix only fixes one compiler_crasher.
However an upcoming patch adds a new call to gatherAllSubstitutions()
which caused some crashers to regress, so I'm going to fix it
properly here.
Sema was dutifully tracking conformances that were "used" as part of
type checking, so it could make sure that those conformances got
completed for SILGen to use. However, this information never actually
made it to SILGen, which included its own (more conservative, not
broad enough) heuristics for finding "used" conformances. Teach Sema
to record conformances within the appropriate source file, and have
SILGen reference the conformances when it emits SIL for the source
file.
We did this for func decls in script, so that DI can flag func decls that access script globals before they've been initialized, but we failed to do so for closures, causing us to miss DI violations when closures referenced script globals before their initialization. Fixes rdar://problem/24357063.
...instead of trying to guess it ourselves.
My previous attempt at this (part of the optional pointers work,
bc83940) made a critical mistake because our only test case /also/
referenced UIApplicationMain directly. I've made the test case test
several more situations, and also added what /would/ be an
execution test if our simulator testing handled UI-based tests.
rdar://problem/25712303
Implements SE-0055: https://github.com/apple/swift-evolution/blob/master/proposals/0055-optional-unsafe-pointers.md
- Add NULL as an extra inhabitant of Builtin.RawPointer (currently
hardcoded to 0 rather than being target-dependent).
- Import non-object pointers as Optional/IUO when nullable/null_unspecified
(like everything else).
- Change the type checker's *-to-pointer conversions to handle a layer of
optional.
- Use 'AutoreleasingUnsafeMutablePointer<NSError?>?' as the type of error
parameters exported to Objective-C.
- Drop NilLiteralConvertible conformance for all pointer types.
- Update the standard library and then all the tests.
I've decided to leave this commit only updating existing tests; any new
tests will come in the following commits. (That may mean some additional
implementation work to follow.)
The other major piece that's missing here is migration. I'm hoping we get
a lot of that with Swift 1.1's work for optional object references, but
I still need to investigate.
Similarly to how we've always handled parameter types, we
now recursively expand tuples in result types and separately
determine a result convention for each result.
The most important code-generation change here is that
indirect results are now returned separately from each
other and from any direct results. It is generally far
better, when receiving an indirect result, to receive it
as an independent result; the caller is much more likely
to be able to directly receive the result in the address
they want to initialize, rather than having to receive it
in temporary memory and then copy parts of it into the
target.
The most important conceptual change here that clients and
producers of SIL must be aware of is the new distinction
between a SILFunctionType's *parameters* and its *argument
list*. The former is just the formal parameters, derived
purely from the parameter types of the original function;
indirect results are no longer in this list. The latter
includes the indirect result arguments; as always, all
the indirect results strictly precede the parameters.
Apply instructions and entry block arguments follow the
argument list, not the parameter list.
A relatively minor change is that there can now be multiple
direct results, each with its own result convention.
This is a minor change because I've chosen to leave
return instructions as taking a single operand and
apply instructions as producing a single result; when
the type describes multiple results, they are implicitly
bound up in a tuple. It might make sense to split these
up and allow e.g. return instructions to take a list
of operands; however, it's not clear what to do on the
caller side, and this would be a major change that can
be separated out from this already over-large patch.
Unsurprisingly, the most invasive changes here are in
SILGen; this requires substantial reworking of both call
emission and reabstraction. It also proved important
to switch several SILGen operations over to work with
RValue instead of ManagedValue, since otherwise they
would be forced to spuriously "implode" buffers.
There's a group of methods in `DeclContext` with names that start with *is*,
such as `isClassOrClassExtensionContext()`. These names suggests a boolean
return value, while the methods actually return a type declaration. This
patch replaces the *is* prefix with *getAs* to better reflect their interface.
_BridgedNSError conformances can affect the runtime behavior of
dynamic casts (e.g. 'is'). Unfortunately, the conformance is not
always emitted, in an effort to save space when not used. This change
forces the conformance witness tables to be emitted when we can detect
a dynamic cast to an _BridgedNSError conforming enum.
Test cases included, as well as a note about a potentially erroneous
path that is not currently handled: when the dynamic cast occurs in a
generic function to a generic type (and thus we are unsure which
conformances we need to pull in).
As there are no instructions left which produce multiple result values, this is a NFC regarding the generated SIL and generated code.
Although this commit is large, most changes are straightforward adoptions to the changes in the ValueBase and SILValue classes.
And use the new project_existential_box to get to the address value.
SILGen now generates a project_existential_box for each alloc_existential_box.
And IRGen re-uses the address value from the alloc_existential_box if the operand of project_existential_box is an alloc_existential_box.
This lets the generated code be the same as before.
And use project_box to get to the address value.
SILGen now generates a project_box for each alloc_box.
And IRGen re-uses the address value from the alloc_box if the operand of project_box is an alloc_box.
This lets the generated code be the same as before.
Other than that most changes of this (quite large) commit are straightforward.
The main idea here is that we really, really want to be
able to recover the protocol requirement of a conformance
reference even if it's abstract due to the conforming type
being abstract (e.g. an archetype). I've made the conversion
from ProtocolConformance* explicit to discourage casual
contamination of the Ref with a null value.
As part of this change, always make conformance arrays in
Substitutions fully parallel to the requirements, as opposed
to occasionally being empty when the conformances are abstract.
As another part of this, I've tried to proactively fix
prospective bugs with partially-concrete conformances, which I
believe can happen with concretely-bound archetypes.
In addition to just giving us stronger invariants, this is
progress towards the removal of the archetype from Substitution.
Parameters (to methods, initializers, accessors, subscripts, etc) have always been represented
as Pattern's (of a particular sort), stemming from an early design direction that was abandoned.
Being built on top of patterns leads to patterns being overly complicated (e.g. tuple patterns
have to have varargs and default parameters) and make working on parameter lists complicated
and error prone. This might have been ok in 2015, but there is no way we can live like this in
2016.
Instead of using Patterns, carve out a new ParameterList and Parameter type to represent all the
parameter specific stuff. This simplifies many things and allows a lot of simplifications.
Unfortunately, I wasn't able to do this very incrementally, so this is a huge patch. The good
news is that it erases a ton of code, and the technical debt that went with it. Ignoring test
suite changes, we have:
77 files changed, 2359 insertions(+), 3221 deletions(-)
This patch also makes a bunch of wierd things dead, but I'll sweep those out in follow-on
patches.
Fixes <rdar://problem/22846558> No code completions in Foo( when Foo has error type
Fixes <rdar://problem/24026538> Slight regression in generated header, which I filed to go with 3a23d75.
Fixes an overloading bug involving default arguments and curried functions (see the diff to
Constraints/diagnostics.swift, which we now correctly accept).
Fixes cases where problems with parameters would get emitted multiple times, e.g. in the
test/Parse/subscripting.swift testcase.
The source range for ParamDecl now includes its type, which permutes some of the IDE / SourceModel tests
(for the better, I think).
Eliminates the bogus "type annotation missing in pattern" error message when a type isn't
specified for a parameter (see test/decl/func/functions.swift).
This now consistently parenthesizes argument lists in function types, which leads to many diffs in the
SILGen tests among others.
This does break the "sibling indentation" test in SourceKit/CodeFormat/indent-sibling.swift, and
I haven't been able to figure it out. Given that this is experimental functionality anyway,
I'm just XFAILing the test for now. i'll look at it separately from this mongo diff.
When enabled, generate closure functions with guaranteed conventions as their context parameters, and pass context arguments to them as guaranteed when possible. (When forming a closure by partial_apply, the partial apply still needs to take ownership of the parameters, regardless of their convention.)
Move these to SILDeclRef, maybe not the best place but a good home for now.
Factor out a new requiresForeignToNativeThunk() function, which cleans up
some code duplication introduced by the following patch:
478e1c7513
This is a small step towards consolidating duplicated logic for figuring out
method dispatch semantics and emitting curry thunks.
Now that we open-code enum construction, enum constructor entry points are
only needed when they are partially-applied, which is a rare case. So we
treat them like curry thunks and only emit them as needed.
The main consequence of this is that enum case constructors are no longer
part of our ABI.
To avoid a regression in the code path for diagnosing infinite value types,
force type lowering to walk a type when emitting its declaration, even if
there are no other references to the type in the program (which is now the
case for public enums which are otherwise not used).
Also XFAIL a DebugInfo test since it is not clear to me what the test does
or how to fix it. The obvious change of adding references to the enum
case constructor function to force it to be emitted did not work.
There's no longer a direct use of the variable's address when we invoke the closure, but we have a handy mark_function_escape instruction to mark the use without requiring merging analysis of the box and its contents. This also gives us a slightly more accurate error message when a variable is prematurely captured, noting the variable was captured by a closure instead of just generically used.
Now that boxes are typed and projectable, the address no longer has to be passed separately.
For now, this breaks capture promotion, DI, and debug info, which analyze uses of the address param. Will be addressed in upcoming commits:
Swift :: DebugInfo/byref-capture.swift
Swift :: DebugInfo/closure-args.swift
Swift :: DebugInfo/closure-args2.swift
Swift :: DebugInfo/inout.swift
Swift :: DebugInfo/linetable.swift
Swift :: SILPasses/capture_promotion.swift
Swift :: SILPasses/definite_init_diagnostics.swift
The drivers for this change are providing a simpler API to SIL pass
authors, having a more efficient of the in-memory representation,
and ruling out an entire class of common bugs that usually result
in hard-to-debug backend crashes.
Summary
-------
SILInstruction
Old New
+---------------+ +------------------+ +-----------------+
|SILInstruction | |SILInstruction | |SILDebugLocation |
+---------------+ +------------------+ +-----------------+
| ... | | ... | | ... |
|SILLocation | |SILDebugLocation *| -> |SILLocation |
|SILDebugScope *| +------------------+ |SILDebugScope * |
+---------------+ +-----------------+
We’re introducing a new class SILDebugLocation which represents the
combination of a SILLocation and a SILDebugScope.
Instead of storing an inline SILLocation and a SILDebugScope pointer,
SILInstruction now only has one SILDebugLocation pointer. The APIs of
SILBuilder and SILDebugLocation guarantees that every SILInstruction
has a nonempty SILDebugScope.
Developer-visible changes include:
SILBuilder
----------
In the old design SILBuilder populated the InsertedInstrs list to
allow setting the debug scopes of all built instructions in bulk
at the very end (as the responsibility of the user). In the new design,
SILBuilder now carries a "current debug scope" state and immediately
sets the debug scope when an instruction is inserted.
This fixes a use-after-free issue with with SIL passes that delete
instructions before destroying the SILBuilder that created them.
Because of this, SILBuilderWithScopes no longer needs to be a template,
which simplifies its call sites.
SILInstruction
--------------
It is neither possible or necessary to manually call setDebugScope()
on a SILInstruction any more. The function still exists as a private
method, but is only used when splicing instructions from one function
to another.
Efficiency
----------
In addition to dropping 20 bytes from each SILInstruction,
SILDebugLocations are now allocated in the SILModule's bump pointer
allocator and are uniqued by SILBuilder. Unfortunately repeat compiles
of the standard library already vary by about 5% so I couldn’t yet
produce reliable numbers for how much this saves overall.
rdar://problem/22017421
- If a @convention(block) function parameter was also marked @noescape, then during type-checking, we would accidentally propagate the convention directly onto a literal closure expr, instead of going through a function_conversion, which SILGen didn't handle. Fixes rdar://problem/23261912.
- If an Objective-C API declared a block parameter with a _Nonnull return of a bridged type, such as NSString *_Nonnull, then native-to-bridged thunking would fail to recognize this case, since we still bridge to an Optional type in the lowered ObjC interface. Fixes rdar://problem/23285766.
