Relax some existing pattern matches and add some unhandled instructions to the
walkers so that borrowing switches over address-only enums are properly analyzed
for incorrect consumption. Add a `[strict]` flag to `mark_unresolved_move_only_value`
to indicate a borrow access that should remain a borrow access even if the subject
is later stack-promoted from a box.
In preparation for inserting mark_dependence instructions for lifetime
dependencies early, immediately after SILGen. That will simplify the
implementation of borrowed arguments.
Marking them unresolved is needed to make OSSA verification
conservative until lifetime dependence diagnostics runs.
This adds SIL-level support and LLVM codegen for normal results of a coroutine.
The main user of this will be autodiff as VJP of a coroutine must be a coroutine itself (in order to produce the yielded result) and return a pullback closure as a normal result.
For now only direct results are supported, but this seems to be enough for autodiff purposes.
We already had getValue() variants... this just expands those to operands. The
reason why this is useful is that these instructions have specific operands with
specific meanings as assigned by an internal enum struct. Rather than having to
access those directly by using getAllOperand()[EnumCase]... better to have a
named method.
Even if the final pattern ends up consuming the value, the match itself
must be nondestructive, because any match condition could fail and cause
us to have to go back to the original aggregate. For copyable values,
we can always copy our way out of consuming operations, but we don't
have that luxury for noncopyable types, so the entire match operation
has to be done as a borrow.
For address-only enums, this requires codifying part of our tag layout
algorithm in SIL, namely that an address-only enum will never use
spare bits or other overlapping storage for the enum tag. This allows
us to assume that `unchecked_take_enum_data_addr` is safely non-side-
effecting and match an address-only noncopyable enum as a borrow.
I put TODOs to remove defensive copies from various parts of our
copyable enum codegen, as well as to have the instruction report
its memory behavior as `None` when the projection is nondestructive,
but this disturbs SILGen for existing code in ways SIL passes aren't
yet ready for, so I'll leave those as is for now.
This patch is enough to get simple examples of noncopyable enum switches
to SILGen correctly. Additional work is necessary to stage in the binding
step of the pattern match; for a consuming switch, we'll need to end
the borrow(s) and then reproject the matched components so we can
consume them moving them into the owned bindings. The move-only checker
also needs to be updated because it currently always tries to convert
a switch into a consuming operation.
NOTE: This does not handle yet assignment into transferring parameters. In the
next commit, I am going to teach the checker that assigning into such a
parameter is a transfer.
Optionally, the dependency to the initialization of the global can be specified with a dependency token `depends_on <token>`.
This is usually a `builtin "once"` which calls the initializer for the global variable.
The dependent 'value' may be marked 'nonescaping', which guarantees that the
lifetime dependence is statically enforceable. In this case, the compiler
must be able to follow all values forwarded from the dependent 'value', and
recognize all final (non-forwarded, non-escaping) use points. This implies
that `findPointerEscape` is false. A diagnostic pass checks that the
incoming SIL to verify that these use points are all initially within the
'base' lifetime. Regular 'mark_dependence' semantics ensure that
optimizations cannot violate the lifetime dependence after diagnostics.
Previously, mayRequirePackMetadata only considered whether a type
involved a pack. That failed to account for the case of outlined value
functions that require pack metadata when the type involves a pack in
its layout. Here, mayRequirePackMetadata now considers also whether the
layout corresponding to a type involves a pack.
rdar://119829826
Some notes:
This is not emitted by SILGen. This is just intended to be used so I can write
SIL test cases for transfer non sendable. I did this by adding an
ActorIsolationCrossing field to all FullApplySites rather than adding it into
the type system on a callee. The reason that this makes sense from a modeling
perspective is that an actor isolation crossing is a caller concept since it is
describing a difference in between the caller's and callee's isolation. As a
bonus it makes this a less viral change.
For simplicity, I made it so that the isolation is represented as an optional
modifier on the instructions:
apply [callee_isolation=XXXX] [caller_isolation=XXXX]
where XXXX is a printed representation of the actor isolation.
When neither callee or caller isolation is specified then the
ApplyIsolationCrossing is std::nullopt. If only one is specified, we make the
other one ActorIsolation::Unspecified.
This required me to move ActorIsolationCrossing from AST/Expr.h ->
AST/ActorIsolation.h to work around compilation issues... Arguably that is where
it should exist anyways so it made sense.
rdar://118521597
* `alloc_vector`: allocates an uninitialized vector of elements on the stack or in a statically initialized global
* `vector`: creates an initialized vector in a statically initialized global
This is another NFC refactor in preparation for changing how we emit
errors. Specifically, we need access to not only the instruction, but also the
specific operand that the transfer occurs at. This ensures that we can look up
the specific type information later when we emit an error rather than tracking
this information throughout the entire pass.
I also included changes to the rest of the SIL optimizer pipeline to ensure that
the part of the optimizer pipeline before we lower tuple_addr_constructor (which
is right after we run TransferNonSendable) work as before.
The reason why I am doing this is that this ensures that diagnostic passes can
tell the difference in between:
```
x = (a, b, c)
```
and
```
x.0 = a
x.1 = b
x.2 = c
```
This is important for things like TransferNonSendable where assigning over the
entire tuple element is treated differently from if one were to initialize it in
pieces using projections.
rdar://117880194
This commit just introduces the instruction. In a subsequent commit, I am going
to add support to SILGen to emit this. This ensures that when we assign into a
tuple var we initialize it with one instruction instead of doing it in pieces.
The problem with doing it in pieces is that when one is emitting diagnostics it
looks semantically like SILGen actually is emitting code for initializing in
pieces which could be an error.
This instruction was given forwarding ownership in the original OSSA
implementation. That will obviously lead to memory leaks. Remove
ownership from this instruction and verify that it is never used for
non-trivial types.
In C++20, the compiler will synthesize a version of the operator
with its arguments reversed to ease commutativity. This reversed
version is ambiguous with the hand-written operator when the
argument is const but `this` isn't.
For instructions at the back of a block, visiting the subsequent
instructions means visiting the instructions at the front of every
successor. For instructions at the front of a block, visiting the prior
instructions means visiting the instructions at the back of every
predecessor.
This instructions marks the point where all let-fields of a class are initialized.
This is important to ensure the correctness of ``ref_element_addr [immutable]`` for let-fields,
because in the initializer of a class, its let-fields are not immutable, yet.
Codegen is the same, but `begin_dealloc_ref` consumes the operand and produces a new SSA value.
This cleanly splits the liferange to the region before and within the destructor of a class.
I was originally hoping to reuse mark_must_check for multiple types of checkers.
In practice, this is not what happened... so giving it a name specifically to do
with non copyable types makes more sense and makes the code clearer.
Just a pure rename.
