LLVM will eventually switch over to using global-isel on arm64 archs.
Setting this option (SWIFT_ENABLE_GLOBAL_ISEL_ARM64) can be used to experiment
with that in Swift before the switch happens.
Our custom build code already sets the architecture of compilation
commands by different means, and so far we were relying on setting
`CMAKE_OSX_ARCHITECTURES` to an empty value to avoid CMake doing the
same.
However, on Apple Silicon, CMake 3.19+ enforces a default value in this
scenario
(https://gitlab.kitware.com/cmake/cmake/-/merge_requests/5291/diffs),
and this would result in the inability to compile code for x86_64.
Addresses SR-14035, rdar://80613594
We have two directories for Swift libraries,
* `SWIFT_SDK_<platform>_LIB_SUBDIR`, a.k.a., the SDK subdir,
a.k.a., `swift-<platform>-<arch>/lib/swift/<platform>`, and
* `SWIFTLIB_SINGLE_SUBDIR`,
a.k.a., `swift-<platform>-<arch>/lib/swift/<platform>/<arch>`.
Through the Swift build, libraries are emitted to both
`.../lib/swift/<platform>` and `.../lib/swift/<platform>/<arch>`.
However, when building toolchains, only `.../lib/swift/<platform>/` is
populated with libraries.
None of this normally isn't a problem; the Swift libraries do not have
inherent interdependencies. This however changes with Concurrency:
Concurrency has an implicit dependency on libdispatch.
When Swift is built, we have two copies of `libswift_Concurrency.so`:
one in `.../lib/swift/<platform>` and one in
`.../lib/swift/<platform>/<arch>`. Prior to this commit, we
unconditionally copy `libdispatch.so` and `libBlocksRuntime.so` to only
_one_ place -- that is, `.../lib/swift/<platform>/<arch>`.
swiftc emits binaries on ELF systems with an rpath of
`.../lib/swift/<platform>`. These binaries implicitly import
Concurrency, so they link against `libswift_Concurrency.so` (whether
they use Concurrency features or not). The library's `$ORIGIN` is
searched to find `libdispatch.so`.
Now, nothing breaks on Linux because there the loader, when given an
rpath, searches both `.../lib/swift/<platform>` and
`.../lib/swift/<platform>/<arch>` even though the rpath only specifies
one directory..
However, on other platforms, only the given rpath is searched.
`libdispatch.so` does not reside next to `libswift_Concurrency.so`
because it has been copied to `.../lib/swift/<platform>/<arch>`; not in
the rpath.
There are a few ways to solve this: change the way rpaths are
configured, only emit libraries into one place, copy `libdispatch.so`
only to the path matching the rpath, or copy `libdispatch.so` wherever
`libswift_Concurrency.so` is copied,
Because the toolchain file layout is different to the file layout when
only Swift is built, hacking the rpath is brittle. Presumably, the
reason why we have a `libswift_Concurrency.so` residing in two places is
to support builds where multiple architectures are supported in the one
build directory, so we cannot just copy `libdispatch.so` _only_ to
`.../lib/swift/<platform>`.
Ultimately, We need to ensure that every instance where
`libswift_Concurrency.so` can be used has `libdispatch.so` residing next
to it, which we do here.
Note that this implicit dependency resolution would not happen unless we
added a `-ldispatch` flag to make this all work, but other platforms are
instaed using `$ORIGIN` to get the search to work properly, so we also
do this for OpenBSD in this commit.
* [Distributed] Initial distributed checking
* [Distributed] initial types shapes and conform to DistributedActor
* [Distributed] Require Codable params and return types
* [Distributed] initial synthesis of fields and constructors
* [Distributed] Field and initializer synthesis
* [Distributed] Codable requirement on distributed funcs; also handle <T: Codable>
* [Distributed] handle generic type params which are Codable in dist func
[Distributed] conformsToProtocol after all
* [Distributed] Implement remote flag on actors
* Implement remote flag on actors
* add test
* actor initializer that sets remote flag
[Distributed] conformances getting there
* [Distributed] dont require async throws; cleanup compile tests
* [Distributed] do not synthesize default implicit init, only our special ones
* [Distributed] properly synth inits and properties; mark actorTransport as _distributedActorIndependent
Also:
- do not synthesize default init() initializer for dist actor
* [Distributed] init(transport:) designated and typechecking
* [Distributed] dist actor initializers MUST delegate to local-init
* [Distributed] check if any ctors in delegation call init(transport:)
* [Distributed] check init(transport:) delegation through many inits; ban invoking init(resolve:using:) explicitly
* [Distributed] disable IRGen test for now
* [Distributed] Rebase cleanups
* [Concurrent] transport and address are concurrent value
* [Distributed] introduce -enable-experimental-distributed flag
* rebase adjustments again
* rebase again...
* [Distributed] distributed functions are implicitly async+throws outside the actor
* [Distributed] implicitly throwing and async distributed funcs
* remove printlns
* add more checks to implicit function test
* [Distributed] resolve initializer now marks the isRemote actor flag
* [Distributed] distributedActor_destroy invoked instead, rather than before normal
* [Distributed] Generate distributed thunk for actors
* [distributed] typechecking for _remote_ functions existing, add tests for remote funcs
* adding one XFAIL'ed task & actor lifetime test
The `executor_deinit1` test fails 100% of the time
(from what I've seen) so I thought we could track
and see when/if someone happens to fix this bug.
Also, added extra coverage for #36298 via `executor_deinit2`
* Fix a memory issue with actors in the runtime system, by @phausler
* add new test that now passes because of patch by @phausler
See previous commit in this PR.
Test is based on one from rdar://74281361
* fix all tests that require the _remote_ function stubs
* Do not infer @actorIndependent onto `let` decls
* REVERT_ME: remove some tests that hacky workarounds will fail
* another flaky test, help build toolchain
* [Distributed] experimental distributed implies experimental concurrency
* [Distributed] Allow distributed function that are not marked async or throws
* [Distributed] make attrs SIMPLE to get serialization generated
* [Distributed] ActorAddress must be Hashable
* [Distributed] Implement transport.actorReady call in local init
* cleanup after rebase
* [Distributed] add availability attributes to all distributed actor code
* cleanup - this fixed some things
* fixing up
* fixing up
* [Distributed] introduce new Distributed module
* [Distributed] diagnose when missing 'import _Distributed'
* [Distributed] make all tests import the module
* more docs on address
* [Distributed] fixup merge issues
* cleanup: remove unnecessary code for now SIMPLE attribute
* fix: fix getActorIsolationOfContext
* [Distributed] cmake: depend on _concurrency module
* fixing tests...
* Revert "another flaky test, help build toolchain"
This reverts commit 83ae6654dd.
* remove xfail
* clenup some IR and SIL tests
* cleanup
* [Distributed] fix cmake test and ScanDependencies/can_import_with_map.swift
* [Distributed] fix flags/build tests
* cleanup: use isDistributed wherever possible
* [Distributed] don't import Dispatch in tests
* dont link distributed in stdlib unittest
* trying always append distributed module
* cleanups
* [Distributed] move all tests to Distributed/ directory
* [lit] try to fix lit test discovery
* [Distributed] update tests after diagnostics for implicit async changed
* [Distributed] Disable remote func tests on Windows for now
* Review cleanups
* [Distributed] fix typo, fixes Concurrency/actor_isolation_objc.swift
* [Distributed] attributes are DistributedOnly (only)
* cleanup
* [Distributed] cleanup: rely on DistributedOnly for guarding the keyword
* Update include/swift/AST/ActorIsolation.h
Co-authored-by: Doug Gregor <dgregor@apple.com>
* introduce isAnyThunk, minor cleanup
* wip
* [Distributed] move some type checking to TypeCheckDistributed.cpp
* [TypeCheckAttr] remove extra debug info
* [Distributed/AutoDiff] fix SILDeclRef creation which caused AutoDiff issue
* cleanups
* [lit] remove json import from lit test suite, not needed after all
* [Distributed] distributed functions only in DistributedActor protocols
* [Distributed] fix flag overlap & build setting
* [Distributed] Simplify noteIsolatedActorMember to not take bool distributed param
* [Distributed] make __isRemote not public
* [Distributed] Fix availability and remove actor class tests
* [actorIndependent] do not apply actorIndependent implicitly to values where it would be illegal to apply
* [Distributed] disable tests until issue fixed
Co-authored-by: Dario Rexin <drexin@apple.com>
Co-authored-by: Kavon Farvardin <kfarvardin@apple.com>
Co-authored-by: Doug Gregor <dgregor@apple.com>
This mimicks the same logic used for shared libraries a few lines above.
This will allow the minimal freestanding stdlib to build with #37511, solving errors like
```
ninja: error: build.ninja:4213: multiple rules generate stdlib/public/SwiftOnoneSupport/CMakeFiles/
swiftSwiftOnoneSupport-freestanding-x86_64-static.dir/__/__/
linker-support/magic-symbols-for-install-name.c.o [-w dupbuild=err]
```
Addresses rdar://78336832
The actual XCTest overlay has not been maintained in this repository for years. (It is distributed in Xcode (libXCTestSwiftSupport.dylib), along with XCTest.framework itself.)
The code we have here is badly out of date, and the fact that we have it on the module path in CI makes it interfere with the contents of recent SDKs.
Remove the XCTest overlay from this repository; pick it up from the SDK instead.
rdar://76915582
* Revert "[Distributed] disable tests until issue fixed"
This reverts commit 0a04278920.
* Revert "[Distributed] Initial `distributed` actors and functions and new module (#37109)"
This reverts commit 814ede0cf3.
* [Distributed] Initial distributed checking
* [Distributed] initial types shapes and conform to DistributedActor
* [Distributed] Require Codable params and return types
* [Distributed] initial synthesis of fields and constructors
* [Distributed] Field and initializer synthesis
* [Distributed] Codable requirement on distributed funcs; also handle <T: Codable>
* [Distributed] handle generic type params which are Codable in dist func
[Distributed] conformsToProtocol after all
* [Distributed] Implement remote flag on actors
* Implement remote flag on actors
* add test
* actor initializer that sets remote flag
[Distributed] conformances getting there
* [Distributed] dont require async throws; cleanup compile tests
* [Distributed] do not synthesize default implicit init, only our special ones
* [Distributed] properly synth inits and properties; mark actorTransport as _distributedActorIndependent
Also:
- do not synthesize default init() initializer for dist actor
* [Distributed] init(transport:) designated and typechecking
* [Distributed] dist actor initializers MUST delegate to local-init
* [Distributed] check if any ctors in delegation call init(transport:)
* [Distributed] check init(transport:) delegation through many inits; ban invoking init(resolve:using:) explicitly
* [Distributed] disable IRGen test for now
* [Distributed] Rebase cleanups
* [Concurrent] transport and address are concurrent value
* [Distributed] introduce -enable-experimental-distributed flag
* rebase adjustments again
* rebase again...
* [Distributed] distributed functions are implicitly async+throws outside the actor
* [Distributed] implicitly throwing and async distributed funcs
* remove printlns
* add more checks to implicit function test
* [Distributed] resolve initializer now marks the isRemote actor flag
* [Distributed] distributedActor_destroy invoked instead, rather than before normal
* [Distributed] Generate distributed thunk for actors
* [distributed] typechecking for _remote_ functions existing, add tests for remote funcs
* adding one XFAIL'ed task & actor lifetime test
The `executor_deinit1` test fails 100% of the time
(from what I've seen) so I thought we could track
and see when/if someone happens to fix this bug.
Also, added extra coverage for #36298 via `executor_deinit2`
* Fix a memory issue with actors in the runtime system, by @phausler
* add new test that now passes because of patch by @phausler
See previous commit in this PR.
Test is based on one from rdar://74281361
* fix all tests that require the _remote_ function stubs
* Do not infer @actorIndependent onto `let` decls
* REVERT_ME: remove some tests that hacky workarounds will fail
* another flaky test, help build toolchain
* [Distributed] experimental distributed implies experimental concurrency
* [Distributed] Allow distributed function that are not marked async or throws
* [Distributed] make attrs SIMPLE to get serialization generated
* [Distributed] ActorAddress must be Hashable
* [Distributed] Implement transport.actorReady call in local init
* cleanup after rebase
* [Distributed] add availability attributes to all distributed actor code
* cleanup - this fixed some things
* fixing up
* fixing up
* [Distributed] introduce new Distributed module
* [Distributed] diagnose when missing 'import _Distributed'
* [Distributed] make all tests import the module
* more docs on address
* [Distributed] fixup merge issues
* cleanup: remove unnecessary code for now SIMPLE attribute
* fix: fix getActorIsolationOfContext
* [Distributed] cmake: depend on _concurrency module
* fixing tests...
* Revert "another flaky test, help build toolchain"
This reverts commit 83ae6654dd.
* remove xfail
* clenup some IR and SIL tests
* cleanup
* [Distributed] fix cmake test and ScanDependencies/can_import_with_map.swift
* [Distributed] fix flags/build tests
* cleanup: use isDistributed wherever possible
* [Distributed] don't import Dispatch in tests
* dont link distributed in stdlib unittest
* trying always append distributed module
* cleanups
* [Distributed] move all tests to Distributed/ directory
* [lit] try to fix lit test discovery
* [Distributed] update tests after diagnostics for implicit async changed
* [Distributed] Disable remote func tests on Windows for now
* Review cleanups
* [Distributed] fix typo, fixes Concurrency/actor_isolation_objc.swift
* [Distributed] attributes are DistributedOnly (only)
* cleanup
* [Distributed] cleanup: rely on DistributedOnly for guarding the keyword
* Update include/swift/AST/ActorIsolation.h
Co-authored-by: Doug Gregor <dgregor@apple.com>
* introduce isAnyThunk, minor cleanup
* wip
* [Distributed] move some type checking to TypeCheckDistributed.cpp
* [TypeCheckAttr] remove extra debug info
* [Distributed/AutoDiff] fix SILDeclRef creation which caused AutoDiff issue
* cleanups
* [lit] remove json import from lit test suite, not needed after all
* [Distributed] distributed functions only in DistributedActor protocols
* [Distributed] fix flag overlap & build setting
* [Distributed] Simplify noteIsolatedActorMember to not take bool distributed param
* [Distributed] make __isRemote not public
Co-authored-by: Dario Rexin <drexin@apple.com>
Co-authored-by: Kavon Farvardin <kfarvardin@apple.com>
Co-authored-by: Doug Gregor <dgregor@apple.com>
Since the NDK removes the platforms/ and sysroot/ directories in the latest NDK
22, switch to the unified sysroot in toolchains/llvm/ and take advantage of a
bunch of simplification that's now possible.
There's no reason to use -m${platform}-version-min as of clang-11/Xcode 11. Clang is now smart enough to parse -target and provide Apple's ld with the appropriate -platform_version argument string.
Credit for the cmake fix here goes to Saleem Abdulrasool.
The substantive fix is embarrassing; I didn't pay close attention
to the intrinsic's argument order and just assumed that the first
argument for the replacement value was the low half (the part
you'd find at index 0 if it were an array), but in fact it's the
high half (the part you'd find at index 1).
I also change the code to be much more reinterpret_casty, which
isolates the type-punning mostly "within" the intrinsic, and
which seems to match how other code uses it.
We already get warnings from clang at compile time. For cases where we silence those, with SWIFT_ALLOWED_RUNTIME_GLOBAL_CTOR_BEGIN, these warnings are just noise.
rdar://problem/70445131
This fixes issues when building Swift with concurrency enabled by default. See rdar://71045297
If some of our overlays need to use the concurrency API, they can manually import _Concurrency.
- deduplicate the logic to compute the resource folder
- install headers and module files in shared and static resource folders
- forward -static flag when calling swiftc with -print-target-info
Since libDemangling is included in the Swift standard library,
ODR violations can occur on platforms that allow statically
linking stdlib if Swift code is linked with other compiler
libraries that also transitively pull in libDemangling, and if
the stdlib version and compiler version do not match exactly
(even down to commit drift between releases). This lets the
runtime conditionally segregate its copies of the libDemangling
symbols from those in the compiler using an inline namespace
without affecting usage throughout source.
The host platform should be using `CMAKE_SYSTEM_NAME STREQUAL Darwin`.
However, we currently drive the host side of the compilation against
custom variables. This makes the migration simpler by ensuring that the
entire file uses the same pattern.
Since `is_darwin_based_sdk` is now used only in the standard library
build, sink it to the standard library build.
Also remove some ancient logic to detect and ignore requests to use LLD.
If people want to explicitly use LLD, they probably have a reason and we
shouldn't second guess them.
Windows had moved to C++14 much earlier than the rest of Swift. This
was required in order to deal with the Windows SDK headers. Now that
the project has fully moved to C++14, remove the unnecessary settings
for Windows.
