This commit adds initial build system support for macCatalyst,
an Apple technology that enables code targeting iOS
to be recompiled so that it can be executed on macOS while still using
iOS APIs. This is the first in a series of commits building out support for
macCatalyst in the compiler, runtime, standard library, and overlays. Swift
for macCatalyst represents the work of multiple people, including
Devin Coughlin, Ross Bayer, and Brent Royal-Gordon.
Under macCatalyst, compiler-provided shared libraries (including overlays)
are built as one of four kinds (or "flavors") of libraries,
each with different install names and Mach-O load commands. This commit
adds the build system infrastructure to produce these different
library flavors.
**macOS-like Libraries**
A "macOS-like" library (such as the GLKit overlay) is a plain-old macOS library
that can only be loaded into regular macOS processes. It has a macOS slice with
a single load command allowing it to be loaded into normal macOS processes.
**iOS-like Libraries**
An "iOS-like" library, such as the UIKit overlay, is a library with a
macOS slice but with a load command that only allows it be loaded into
macCatalyst processes. iOS-like libraries are produced by passing a new
target tuple to the compiler:
swiftc ... -target x86_64-apple-ios13.0-macabi ...
Here 'ios' (and an iOS version number) is used for OS portion
of the triple, but the 'macabi' environment tells the compiler
that the library is intended for macCatalyst.
**Zippered Libraries**
A "zippered" library can be loaded into either a macCatalyst process or
a standard macOS process. Since macCatalyst does not introduce a new Mach-O
slice, the same code is shared between both processes. Zippered libraries
are usually relatively low level and with an API surface that is similar
between macOS and iOS (for example, both the Foundation overlay and the Swift
Standard Library/Runtime itself are zippered).
Zippered libraries are created by passing both the usual `-target`
flag to the compiler and an additional `-target-variant` flag:
swiftc ... -target x86_64-apple-macos10.15 \
-target-variant x86_64-apple-ios13.0-macabi
Just like the -target flag, -target-variant takes a target tuple.
This tells the compiler to compile the library for the -target tuple but
to add an extra load command, allowing the library to be loaded into processes
of the -target-variant flavor as well.
While a single zippered library and slice is shared between macOS and
macCatalyst, zippered libraries require two separate .swiftinterface/.swiftmodule
files, one for macOS and one for macCatalyst. When a macOS or macCatalyst client
imports the library, it will use module file for its flavor to determine what
symbols are present. This enables a zippered library to expose a subset of its
target APIs to its target-variant.
**Unzippered-Twin Libraries**
"Unzippered Twins" are pairs of libraries with the same name but different
contents and install locations, one for use from macOS processes and one for
use from macCatalyst processes. Unzippered twins are usually libraries that
depend on AppKit on macOS and UIKit on iOS (for example, the MapKit overlay)
and so do not share a common implementation between macOS and macCatalyst.
The macCatalyst version of an unzippered twin is installed in a parallel
directory hierarchy rooted at /System/iOSSupport/. So, for example, while macOS
and zippered Swift overlays are installed in /usr/lib/swift/, iOS-like and
the macCatalyst side of unzippered twins are installed in
/System/iOSSupport/usr/lib/swift. When building for macCatalyst, the build system
passes additional search paths so that the macCatalyst version of libraries is
found before macOS versions.
The add_swift_target_library() funciton now take an
optional MACCATALYST_BUILD_FLAVOR, which enables swift libraries to indicate
which flavor of library they are.
Move the special flag handling for the non-atomic runtime into the build
system rather than spreading it across the build system and the helper
utilities.
This commit introduces a CMake target for each component, adds install targets
for them, and switches build-script-impl to use the target `install-components`
for installation. Each of the targets for each component depends on each
of the individual targets and outputs that are associated with the
corresponding swift-component.
This is equivalent to what already exists, because right now install rules are
only generated for components that we want to install. Therefore, this commit
should be an NFC.
This is a resubmission (with modifications) of an earlier change. I originally
committed this but there were problems with some installation rules.
This commit introduces a CMake target for each component, adds install targets
for them, and switches build-script-impl to use the target `install-components`
for installation. Each of the targets for each component depends on each
of the individual targets and outputs that are associated with the
corresponding swift-component.
This is equivalent to what already exists, because right now install rules are
only generated for components that we want to install. Therefore, this commit
should be an NFC.
`StaticBinaryELF.cpp` was listed as an optional source but then unused
by any build target. As a result, it escaped detection that the file
was no longer in use.
There are situations where you want to build against a libc that is out
of tree or that is not the system libc (Or for cross build scenarios).
This is a change for passing the -sdk and include paths for things like
this.
CMake supports the notion of installation components. Right now we have some
custom code for supporting swift components. I think that for installation
purposes, it would be nice to use the CMake component system.
This should be a non-functional change. We should still only be generating
install rules for targets and files in components we want to install, and we
still use the install ninja target to install everything.
Recent Swift uses 2 as the is-Swift bit when running on newer versions, and 1 on older versions. Since it's difficult or impossible to know what we'll be running on at build time, make the selection at runtime.
If we nest a type inside a local context inside a generic type,
we have to look through the local context(s) to find the outer
generic type when stripping off generic arguments.
We don't support nominal types inside generic local context
right now, but this can happen with type aliases.
Once upon a time we tried to use this to limit the visibility of
symbols in the Swift runtime in a way that didn't really make sense.
Dave Z removed it last year in 91823273d2.
No functionality change.
* cmake: Propagate SWIFT_DARWIN_ENABLE_STABLE_ABI_BIT to overlay builds.
* runtime: Clear the correct bit in getROData()
* test/IRGen/objc_class_export.swift: Allow either is-Swift bit.
* test/stdlib/SwiftObjectNSObject.swift: Allow either name for SwiftObject.
The object file extension on Windows is `.obj` rather than `.o`. Ensure
that we get the extension correct when compiling for Windows.
Furthermore, ensure that we install with the correct extension when
installing a cross-compiled image. As we may be using the homegrown
cross-compilation system, we must explicitly handle the extension
ourselves. This allows nearly a 100 additional tests to pass on
Windows.
Previously, the stdlib provided:
- getters for AnyKeyPath and PartialKeyPath, which have remained;
- a getter for KeyPath, which still exists alongside a new read
coroutine; and
- a pair of owned mutable addressors that provided modify-like behavior
for WritableKeyPath and ReferenceWritableKeyPath, which have been
replaced with modify coroutines and augmented with dedicated setters.
SILGen then uses the most efficient accessor available for the access
it's been asked to do: for example, if it's been asked to produce a
borrowed r-value, it uses the read accessor.
Providing a broad spectrum of accessor functions here seems acceptable
because the code-size hit is fixed-size: we don't need to generate
extra code per storage declaration to support more alternatives for
key paths.
Note that this is just the compiler ABI; the implementation is still
basically what it was. That means the implementation of the setters
and the read accessor is pretty far from optimal. But we can improve
the implementation later; we can't improve the ABI.
The coroutine accessors have to be implemented in C++ and used via
hand-rolled declarations in SILGen because it's not currently possible
to declare independent coroutine accessors in Swift.
The key thing here is that all of the underlying code is exactly the same. I
purposely did not debride anything. This is to ensure that I am not touching too
much and increasing the probability of weird errors from occurring. Thus the
exact same code should be executed... just the routing changed.
We were previously only installing the image registration helper in the shared
runtime location. When building with a static runtime, we would look in the
wrong location. Ensure that we install to both locations to repair the static
builds.
Use the reserved spelling for the weak attribute. Move the definition into an
implementation file to avoid multiple definitions of the function from being
emitted. PE/COFF does not support weak symbols, however, the runtime will never
link against LLVMSupport, so provide a single, strong definition of the
function. Mark it as hidden visibility so that we do not expose it outside of
the runtime.
Follow-up to 3ed3774e07. On Apple OSs that don't have the new
Objective-C runtime function 'objc_setHook_getImageName', override the
system definition of 'class_getImageName' by literally rewriting
symbol tables at run time.
Yes, you read that correctly.
The low-level part of this patch was written by Greg Parker, then
simplified and tweaked by me to fit the Swift coding style. Don't try
this at home; it comes with all sorts of caveats and won't actually
work on this year's iOS. (Fortunately we don't need it there, because
that will have the new ObjC entry point.)
The rest of the patch is pretty straightforward: the replacement
implementation calls the code that supports Swift objects (the same
code we use on newer OSs), which then chains back to the original
system implementation of class_getImageName. May we never have to
touch this again.
rdar://problem/41535552
Previously, swiftImageInspectionShared generated one specific library at
`lib/libswiftImageInspectionShared.a` for only the main arch and sdk.
Generic cross compilation and various changes to the build system to get
cross compilation to work will require swiftImageInspectionShared to
generate libraries at the proper subdirectory. Change the outputs to
agree with paths such as `lib/swift/linux/x86_64`
When building with assertions enabled, link the demangle tree dumper into
the runtime and remote mirrors libraries. This makes debugging demangling-related issues a whole lot easier.
* SR-106: New floating-point `description` implementation
This replaces the current implementation of `description` and
`debugDescription` for the standard floating-point types with a new
formatting routine based on a variation of Florian Loitsch' Grisu2
algorithm with changes suggested by Andrysco, Jhala, and Lerner's 2016
paper describing Errol3.
Unlike the earlier code based on `sprintf` with a fixed number of
digits, this version always chooses the optimal number of digits. As
such, we can now use the exact same output for both `description` and
`debugDescription` (except of course that `debugDescription` provides
full detail for NaNs).
The implementation has been extensively commented; people familiar with
Grisu-style algorithms should find the code easy to understand.
This implementation is:
* Fast. It uses only fixed-width integer arithmetic and has constant
memory and time requirements.
* Simple. It is only a little more complex than Loitsch' original
implementation of Grisu2. The digit decomposition logic for double is
less than 300 lines of standard C (half of which is common arithmetic
support routines).
* Always Accurate. Converting the decimal form back to binary (using an
accurate algorithm such as Clinger's) will always yield exactly the
original binary value. For the IEEE 754 formats, the round-trip will
produce exactly the same bit pattern in memory. This is an essential
requirement for JSON serialization, debugging, and logging.
* Always Short. This always selects an accurate result with the minimum
number of decimal digits. (So that `1.0 / 10.0` will always print
`0.1`.)
* Always Close. Among all accurate, short results, this always chooses
the result that is closest to the exact floating-point value. (In case
of an exact tie, it rounds the last digit even.)
This resolves SR-106 and related issues that have complained
about the floating-point `description` properties being inexact.
* Remove duplicate infinity handling
* Use defined(__SIZEOF_INT128__) to detect uint128_t support
* Separate `extracting` the integer part from `clearing` the integer part
The previous code was unnecessarily obfuscated by the attempt to combine
these two operations.
* Use `UINT32_MAX` to mask off 32 bits of a larger integer
* Correct the expected NaN results for 32-bit i386
* Make the C++ exceptions here consistent
Adding a C source file somehow exposed an issue in an unrelated C++ file.
Thanks to Joe Groff for the fix.
* Rename SwiftDtoa to ".cpp"
Having a C file in stdlib/public/runtime causes strange
build failures on Linux in unrelated C++ files.
As a workaround, rename SwiftDtoa.c to .cpp to see
if that avoids the problems.
* Revert "Make the C++ exceptions here consistent"
This reverts commit 6cd5c20566.
Commit 0c42b57962 ("ELF: restructure image metadata registration")
removed the swift_install_in_component lines for both
swiftImageInspectionStatic and swiftImageInspectionShared libraries,
even though only the former library was removed in that change. As a
result, the swiftImageInspectionShared was not being built or
installed. This should fix SR-7038.
rdar://problem/37710244
The dumper method dumps:
1. The container's metadata pointer.
2. A pointer to the container's value.
3. Whether or not said value is stored inline in the container.
This provides a general overview that can be used even when working with SIL
code in the debugger by grabbing a pointer to swift Anys and then calling the
c++ any method upon them.
The verifier is intended to be used in conjunction with ASAN for maximum
effect to catch use-after-frees of existential boxes.
While implementing this I refactored some code from ExistentialTypeMetadata into
methods on OpaqueExistentialContainer. ExistentialTypeMetadata just calls these
methods now instead of implementing the code inline.
The old-style mirrors were the basis for non-custom, reflection based Mirrors. As part of removing old-style mirrors, reflection based Mirrors are reimplemented without them. Reflection.mm is deleted and replaced with ReflectionMirror.mm, which borrows a chunk of the old code but presents a simpler API to the Swift side. ReflectionMirror.swift then uses that API to implement a reflection-based Mirror init.
rdar://problem/20356017
* Remove RegisterPreservingCC. It was unused.
* Remove DefaultCC from the runtime. The distinction between C_CC and DefaultCC
was unused and inconsistently applied. Separate C_CC and DefaultCC are
still present in the compiler.
* Remove function pointer indirection from runtime functions except those
that are used by Instruments. The remaining Instruments interface is
expected to change later due to function pointer liability.
* Remove swift_rt_ wrappers. Function pointers are an ABI liability that we
don't want, and there are better ways to get nonlazy binding if we need it.
The fully custom wrappers were only needed for RegisterPreservingCC and
for optimizing the Instruments function pointers.
Various TypeDecoder clients will depend on having the "bare" nominal
type declaration demangled node for looking up nominal type descriptors,
so move the generic argument-stripping code into TypeDecoder.
Swift class metadata has a bit to distinguish it from non-Swift Objective-C
classes. The stable ABI will use a different bit so that stable Swift and
pre-stable Swift can be distinguished from each other.
No bits are actually changed yet. Enabling the new bit needs to wait for
other coordination such as libobjc.
rdar://35767811
Restructure the COFF metadata handling to use the linker section
grouping to emit section start/stop markers in the appropriate location.
This allows us to lookup the sections statically without having to the
walk the entire image structure.
Introduce a constructor for PE/COFF binaries. This will ensure that the
registration occurs for all modules appropriately. This should resolve
rdar://problem/19045112. The registration should occur prior to
`DllMain` being invoked from `DllMainCRTStartup`.
Restructure the ELF handling to be completely agnostic to the OS.
Rather than usng the loader to query the section information, use the
linker to construct linker tables and synthetic markers for the
beginning and of the table. Save off the values of these pointers and
pass them along through the constructor to the runtime for registration.
This removes the need for the begin/end objects. Remove the special
construction of the begin/end objects through the special assembly
constructs, preferring to do this in C with a bit of inline assembly to
ensure that the section is always allocated.
Remove the special handling for the various targets, the empty object
file can be linked on all the targets.
The new object file has no requirements on the ordering. It needs to
simply be injected into the link.
Name the replacement file `swiftrt.o` mirroring `crt.o` from libc. Merge
the constructor and the definition into a single object file.
This approach is generally more portable, overall simpler to implement,
and more robust.
Thanks to Orlando Bassotto for help analyzing some of the odd behaviours
when switching over.
This adds the swift include path manually to the builds for the stubs
and the runtime. This has no impact for the build currently. However,
adding the additional include directory will enable a standalone build
for the stdlib.
* Revert "[strip -ST] Disable runtime stack trace dumping on Darwin when asserts are disabled."
This reverts commit 6bc28ff1c9.
* Bring back important fixes from the revert of 6bc28ff1c9.
* Change swift::swift_reportError to only print the backtrace in assert builds (swift::warning prints backtrace always).
This commit disables runtime stack trace dumping via dladdr on Darwin when
asserts are disabled.
This stack trace dumping was added as a way to improve the ability to debug the
compiler for compiler developers. This is all well and good but having such a
feature always enabled prevents us from reducing the size of the swift standard
library by eliminating the swift nlist.
rdar://31372220
