There's no guarantee that e.g. pthread_key_t is an integral type. It could
be some kind of struct, or some other thing that isn't valid as a template
argument.
rdar://90776105
Moved all the threading code to one place. Added explicit support for
Darwin, Linux, Pthreads, C11 threads and Win32 threads, including new
implementations of Once for Linux, Pthreads, C11 and Win32.
rdar://90776105
SWIFT_STDLIB_SINGLE_THREADED_RUNTIME is too much of a blunt instrument here.
It covers both the Concurrency runtime and the rest of the runtime, but we'd
like to be able to have e.g. a single-threaded Concurrency runtime while
the rest of the runtime is still thread safe (for instance).
So: rename it to SWIFT_STDLIB_SINGLE_THREADED_CONCURRENCY and make it just
control the Concurrency runtime, then add a SWIFT_STDLIB_THREADING_PACKAGE
setting at the CMake/build-script level, which defines
SWIFT_STDLIB_THREADING_xxx where xxx depends on the chosen threading package.
This is especially useful on systems where there may be a choice of threading
package that you could use.
rdar://90776105
There's no guarantee that e.g. pthread_key_t is an integral type. It could
be some kind of struct, or some other thing that isn't valid as a template
argument.
rdar://90776105
Moved all the threading code to one place. Added explicit support for
Darwin, Linux, Pthreads, C11 threads and Win32 threads, including new
implementations of Once for Linux, Pthreads, C11 and Win32.
rdar://90776105
SWIFT_STDLIB_SINGLE_THREADED_RUNTIME is too much of a blunt instrument here.
It covers both the Concurrency runtime and the rest of the runtime, but we'd
like to be able to have e.g. a single-threaded Concurrency runtime while
the rest of the runtime is still thread safe (for instance).
So: rename it to SWIFT_STDLIB_SINGLE_THREADED_CONCURRENCY and make it just
control the Concurrency runtime, then add a SWIFT_STDLIB_THREADING_PACKAGE
setting at the CMake/build-script level, which defines
SWIFT_STDLIB_THREADING_xxx where xxx depends on the chosen threading package.
This is especially useful on systems where there may be a choice of threading
package that you could use.
rdar://90776105
Change continuation signposts to emit an interval for init/resume.
Fix task_create to take the decoded flags as separate parameters, matching other calls.
Move job_run trace calls into runJobInEstablishedExecutorContext. swift_job_runImpl didn't catch everything.
rdar://92149411
Apply a blanket pass of including `new` for the placement new allocation
and namespacing the call to the global placement new allocator. This
should repair the Android ARMv7 builds.
Change actor destruction to call swift_deallocClassInstance instead of swift_deallocObject. When ObjC interop is enabled, swift_deallocClassInstance will check the "pure swift deallocation" bit and call into the ObjC runtime to destruct the instance when needed. This is what clears weak references and associated objects.
rdar://91270492
Not all targets have a 16-byte type alignment guarantee. For the types
which are not naturally aligned, provide a type specific `operator new`
overload to ensure that we are properly aligning the type on allocation
as we run the risk of under-aligned allocations otherwise.
This should no longer be needed with C++17 and newer which do a two
phase `operator new` lookup preferring
`operator new(std::size, std::align_val_t)` if needed. The base type
would be fully pre-processed away. The empty base class optimization
should help ensure that we do not pay any extra size costs for the
alignment fixes.
As we are a C++14 codebase, we must locally implement some of the
standard type_traits utilities, namely `void_t`. We take the minimal
definition here, assuming that the compiler is up-to-date with C++14 DR
reports which fixed an issue in SFINAE. We use the SFINAE for detecting
the presence of the `operator new` overload to guide the over-alignment,
which is inherited through the new `swift::overaligned_type<>` base
type.
Annotate the known classes which request explicit alignment which is
non-pointer alignment. This list was identified by
`git grep ' alignas(.*) '`.
We don't necessarily have environment variables, which is why we have
SWIFT_STDLIB_HAS_ENVIRON. Calling getenv() outside of a check of this
means Concurrency won't work in environments where getenv() cannot
be used.
rdar://91050064
When possible, decode the DrainLock/ExecutionLock fields of tasks and actors in concurrency runtimes built with priority escalation, and show the corresponding thread info in swift-inspect output.
We weren't properly decoding actor flags previously, so fix that up as well and have Remote Mirror split them out into separate fields so clients don't have to. We were missing the Job Storage field from the definition of DefaultActorImpl in RuntimeInternals.h, fix that so we actually read the right data.
rdar://88598003
Decode all fields from the various flags values, pass each field as a separate argument to the various signposts. We were just passing the raw value of the flags and requiring the signpost client to decode them, which was ugly and required the client to know details they shouldn't need to know.
Strip ptrauth bits from the task resume function when signposting, when ptrauth is supported.
Add signpost events for continuation init/await/resume events.
We also make task_wait into an interval, rather than a single event. The interval ends when the task resumes. As part of this change, we also skip emitting the interval when the wait completed immediately and the task didn't have to suspend.
While we're in there, clean up a few SWIFT_TASK_DEBUG_LOG lines that emitted warnings when built with the logging enabled.
rdar://88658803
We cannot access the executed job after it has finished executing:
- If it's a non-task job, it is always invalidated; such jobs are
self-owning, and they are expected to destroy themselves after
execution.
- If it's a task, and it completes during execution, it will invalidate
itself synchronously, e.g. by releasing itself. At this point, it
must be assumed that the task memory has been releaed.
- If it's a task, and it hasn't completed during execution, we are
now racing with whatever event *does* complete the task.
Any information we want to log about the job must be recorded when it
starts to run.
rdar://88817560
preprocessing is due to running into a concurrent enqueuer while
draining. However, sometimes we can go into this code path because
compare_exchange_weak failed spuriously. We need to account the fact
that compare_exchange_weak can fail spuriously. Also add more asserts.
Radar-Id: rdar://problem/89236911
state of the actor similar to the ActiveTaskStatus. Refactor the default
actor runtime implementation to set us up for priority escalation
support
Radar-Id: rdar://problem/86100521
A task can be in one of 4 states over its lifetime:
(a) suspended
(b) enqueued
(c) running
(d) completed
This change provides priority inversion avoidance support if a task gets
escalated when it is in state (a), (c), (d).
Radar-Id: rdar://problem/76127624
This change adds support for WASI in stdlib tests. Some tests that expect a crash to happen had to be disabled, since there's currently no way to observe such crash from a WASI host.
* [Distributed] Adjust interface of `swift_distributed_execute_target`
Since this is a special function, `calleeContext` doesn't point to
a direct parent but instead both parent context (uninitialized)
and resume function are passed as last arguments which means that
`callContext` has to act as an intermediate context in call to accessor.
* [Distributed] Drop optionality from result buffer in `_executeDistributedTarget`
`RawPointer?` is lowered into a two arguments since it's a struct,
to make it easy let's just allocate an empty pointer for `Void` result.
* [Distributed] NFC: Update _remoteCall test-case to check multiple different result types
