The address of the function to be called when generating code to invoke
the function associated with FunctionPointer which is produced via
direct reference is by definition statically known; it is neither necessary
nor desireable to load this address out of the AsyncFunctionPointer
corresponding to the function.
Here, that spurious additional work is skipped. The approach is to add
a second value to the FunctionPointer struct. For FunctionPointers
whose kind is AsyncFunctionPointer, this value is either null or else
the address of the corresponding function.
rdar://71376092
Previously, AsyncFunctionPointer constants were signed as code. That
was incorrect considering that these constants are in fact data. Here,
that is fixed.
rdar://76118522
The comment in LowerHopToActor explains the design here.
We want SILGen to emit hops to actors, ignoring executors,
because it's easier to fully optimize in a world where deriving
an executor is a non-trivial operation. But we also want something
prior to IRGen to lower the executor derivation because there are
useful static optimizations we can do, such as doing the derivation
exactly once on a dominance path and strength-reducing the derivation
(e.g. exploiting static knowledge that an actor is a default actor).
There are probably phase-ordering problems with doing this so late,
but hopefully they're restricted to situations like actors that
share an executor. We'll want to optimize that eventually, but
in the meantime, this unblocks the executor work.
The immediate desire is to minimize the set of ABI dependencies
on the layout of an ExecutorRef. In addition to that, however,
I wanted to generally reduce the code size impact of an unsafe
continuation since it now requires accessing thread-local state,
and I wanted resumption to not have to create unnecessary type
metadata for the value type just to do the initialization.
Therefore, I've introduced a swift_continuation_init function
which handles the default initialization of a continuation
and returns a reference to the current task. I've also moved
the initialization of the normal continuation result into the
caller (out of the runtime), and I've moved the resumption-side
cmpxchg into the runtime (and prior to the task being enqueued).
Throwing functions pass the error result in `swiftself` to the resume
partial function.
Therefore, `() async -> ()` to `() async throws -> ()` is not ABI compatible.
TODO: go through remaining failing IRGen async tests and replace the
illegal convert_functions.
Most of the async runtime functions have been changed to not
expect the task and executor to be passed in. When knowing the
task and executor is necessary, there are runtime functions
available to recover them.
The biggest change I had to make to a runtime function signature
was to swift_task_switch, which has been altered to expect to be
passed the context and resumption function instead of requiring
the caller to park the task. This has the pleasant consequence
of allowing the implementation to very quickly turn around when
it recognizes that the current executor is satisfactory. It does
mean that on arm64e we have to sign the continuation function
pointer as an argument and then potentially resign it when
assigning into the task's resume slot.
rdar://70546948
Previously, thick async functions were represented sometimes as a pair
of (AsyncFunctionPointer, nullptr)--when the thick function was produced
via a thin_to_thick_function, e.g.--and sometimes as a pair of
(FunctionPointer, ThickContext)--when the thick function was produced by
a partial_apply--with the size stored in the slot of the ThickContext.
That optimized for the wrong case: partial applies of dynamic async
functions; in that case, there is no appropriate AsyncFunctionPointer to
form when lowering the partial_apply instruction. The far more common
case is to know exactly which function is being partially applied. In
that case, we can form the appropriate AsyncFunctionPointer.
Furthermore, the previous representation made calling a thick function
more complex: it was always necessary to check whether the context was
in fact null and then proceed along two different paths depending.
Here, that behavior is corrected by creating a thunk in a mandatory
IRGen SIL pass in the case that the function that is being partially
applied is dynamic. That new thunk is then partially applied in place
of the original partial_apply of the dynamic function.
We don't need to initialize the global object if it's never used for something which can access the object header.
This let's e.g. lookup-tables to be compiled with zero Array-overhead.
rdar://59874359
OwnershipEliminator lowers destroy_value [poison] to debug_value
[poison].
IRGen overwrites all references in shadow copies with a sentinel value
in place of debug_value [poison].
Part 2/2: rdar://75012368 (-Onone compiler support for early object
deinitialization with sentinel dead references)
The `coro.end.async` intrinsic allow specifying a function that is to be
tail-called as the last thing before returning.
LLVM lowering will inline the `must-tail-call` function argument to
`coro.end.async`. This `must-tail-call` function can contain a
`musttail` call.
```
define @my_must_tail_call_func(void (*)(i64) %fnptr, i64 %args) {
musttail call void %fnptr(i64 %args)
ret void
}
define @async_func() {
...
coro.end.async(..., @my_must_tail_call_func, %return_continuation, i64 %args)
unreachable
}
```
r11 is the linker scratch register, so we need to make sure that
we don't end up in the linker when making cross-library direct
async calls. Since there are currently only a few special
symbols that we do that for, we can achieve this by disabling
lazy binding for those symbols specifically.
First, just call an async -> T function instead of forcing the caller
to piece together which case we're in and perform its own copy. This
ensures that the task is actually kept alive properly.
Second, now that we no longer implicitly depend on the waiting tasks
being run synchronously, go ahead and schedule them to run on the
global executor.
This solves some problems which were blocking the work on TLS-ifying
the task/executor state.
LLVM seems to not support this today. With ownership SSA, we now produce these
for accelerate for some reason, causing lldb/TestAccelerateSIMD.py to fail. I
debugged this with Adrian and we got this fix. I filed the radar below against
Adrian to fix.
rdar://74287800
Previously, the error stored in the async context was of type SwiftError
*. In order to enable the context to be callee released, make it
indirect and change its type to SwiftError **.
rdar://71378532
LLVM does type based analysis on sret storage types. This is a problem
with non-fixed types whose storage type does not contain the non-fixed
part.
rdar://73778591
Usually debug info only ever describes the *result* of a projectBox
call. To display a boxed parameter of an async continuation object,
however, the debug info can only describe the box itself and thus also
needs to emit a box type for it so the debugger knows to call into
Remote Mirrors to unbox the value.
SR-14059
rdar://problem/73358988
Empty types (such as structs without stored properties) have a
meaningless value for their address. To avoid crashes in the Thread
Sanitizer runtime, rather than passing this unspecified value as
the address of the inout access, skip emission of the runtime call.
The bug allowing unspecified behavior here has been present since we
first added TSan support for checking Swift access races -- but codegen
changes on arm64 have recently made crashes due to the bug much more
likely.
rdar://problem/47686212
This adds new kinds of link entities corresponding to the three
dispatch thunk link entity kinds:
- DispatchThunkAsyncFunctionPointer
- DispatchThunkInitializerAsyncFunctionPointer
- DispatchThunkAllocatorAsyncFunctionPointer
* Initial draft of async sequences
* Adjust AsyncSequence associated type requirements
* Add a draft implementation of AsyncSequence and associated functionality
* Correct merge damage and rename from GeneratorProtocol to AsyncIteratorProtocol
* Add AsyncSequence types to the cmake lists
* Add cancellation support
* [DRAFT] Implementation of protocol conformance rethrowing
* Account for ASTVerifier passes to ensure throwing and by conformance rethrowing verifies appropriately
* Remove commented out code
* OtherConstructorDeclRefExpr can also be a source of a rethrowing kind function
* Re-order the checkApply logic to account for existing throwing calculations better
* Extract rethrowing calculation into smaller functions
* Allow for closures and protocol conformances to contribute to throwing
* Add unit tests for conformance based rethrowing
* Restrict rethrowing requirements to only protocols marked with @rethrows
* Correct logic for gating of `@rethrows` and adjust the determinates to be based upon throws and not rethrows spelling
* Attempt to unify the async sequence features together
* Reorder try await to latest syntax
* revert back to the inout diagnosis
* House mutations in local scope
* Revert "House mutations in local scope"
This reverts commit d91f1b25b59fff8e4be107c808895ff3f293b394.
* Adjust for inout diagnostics and fall back to original mutation strategy
* Convert async flag to source locations and add initial try support to for await in syntax
* Fix case typo of MinMax.swift
* Adjust rethrowing tests to account for changes associated with @rethrows
* Allow parsing and diagnostics associated with try applied to for await in syntax
* Correct the code-completion for @rethrows
* Additional corrections for the code-completion for @rethrows this time for the last in the list
* Handle throwing cases of iteration of async sequences
* restore building XCTest
* First wave of feedback fixes
* Rework constraints checking for async sequence for-try-await-in checking
* Allow testing of for-await-in parsing and silgen testing and add unit tests for both
* Remove async sequence operators for now
* Back out cancellation of AsyncIteratorProtocols
* Restructure protocol conformance throws checking and cache results
* remove some stray whitespaces
* Correct some merge damage
* Ensure the throwing determinate for applying for-await-in always has a valid value and adjust the for-await-in silgen test to reflect the cancel changes
* Squelch the python linter for line length
