`malloc_usable_size` is provided on Android with API Level 17 or newer.
However, the signature between bionic and glibc is slightly different, with
android marking the parameter as `const`. This results in a type mismatch error
when building the standard library for android. Adjust it to match the target
system.
`ssize_t` is not universally available as it is not a standard type.
However, we require clang to build the runtime due to custom calling
convention support. As a result, we know that the compiler will support
the `_Generic` keyword from the C11 standard. Use this with an
expansion to map the type appropriately for all targets. This avoids
having to have a sanity check in the runtime that the type definition
matches the underlying system.
When this header is used to create a module, it is used in a "C" environment,
which means that we cannot use the C++ style casts. Switch to the
old-style-cast. If we need to, we can always use `-Wno-old-style-casts` to
silence the warnings in C++ mode.
Previously we had a single mask for all x86-64 targets which included both the top and bottom bits. This accommodated simulators, which use the top bit, while macOS uses the bottom bit, but reserved one bit more than necessary on each. This change breaks out x86-64 simulators from non-simulators and reserves only the one bit used on each.
rdar://problem/34805348 rdar://problem/29765919
The functions in LibcShims are used externally, some directly and some through @inlineable functions. These are changed to SWIFT_RUNTIME_STDLIB_SPI to better match their actual usage. Their names are also changed to add "_swift" to the front to match our naming conventions.
Three functions from SwiftObject.mm are changed to SPI and get a _swift prefix.
A few other support functions are also changed to SPI. They already had a prefix and look like they were meant to be SPI anyway. It was just hard to notice any mixup when they were #defined to the same thing.
rdar://problem/35863717
- Don’t expose the raw execution seed to _rawHashValue.
- Change the type of _rawHashValue’s seed from (UInt64,UInt64) to a single Int. Working with a pair of UInt64s is unwieldy, and overkill in practice. Int as a seed also integrates nicely with Int as a hash value.
- Remove _HasherCore._generateSeed(). Instead, simply call finalize() on a copy of the hasher to get a seed suitable for _rawHashValue.
- Update Set and Dictionary to store a single Int as the seed value.
Note that this doesn’t affect the core hasher, which still mixes in the actual 128-bit execution seed during its initialization. To reduce the potential of confusion, use the name “rawSeed” to refer to an actual 128-bit seed value.
- Use _HashTable to unify low-level hashing operations across Set and Dictionary.
- Store the capacity directly inside the storage header. This allows the maximum load factor to be controlled by non-inlinable code.
- Introduce a dedicated class for the empty singleton.
- Add _BridgingHashBuffer, a standalone flat hash buffer for use in deferred bridging. Use it to eliminate the need to support non-hashable storage/wrapper variants and to improve memory use in cases where Key or Value are verbatim bridged.
- Eliminate the “TypedNative*Storage” class and _NativeSet/_NativeDictionary’s support for non-hashable keys.
- Rename storage classes as follows:
_RawNativeSetStorage ⟹ _RawSetStorage
_RawNativeDictionaryStorage ⟹ _RawDictionaryStorage
_TypedNativeSetStorage ⟹ (removed)
_TypedNativeDictionaryStorage ⟹ (removed)
_HashableTypedNativeSetStorage ⟹ _SetStorage
_HashableTypedNativeDictionaryStorage ⟹ _DictionaryStorage
The new names make it obvious which ivar layout is in use.
This allows us to move the empty NSSet/NSDictionary overrides out of the root storage class; they don’t really belong there. More importantly, it makes empty sets/dictionaries super obvious to lldb and other runtime tools.
Update the instances of checks for architectures to be more broad for different
spellings of the architecture macro. Certain targets use `_M_ARM` and others
use `__arm__`. Similarly, arm64/aarch64 has `_M_ARM64`, `__arm64__` and
`__aarch64__` as spellings. This just mechanically goes through and encodes the
various spellings.
Take the opportunity to replace some raw checks with `defined` checks which
avoids a pedantic warning due to the undefined macro when performing the check
as the preprocessor may warn about an undefined condition evaluating to `0`.
android on AArch64 will alias `mode_t` to `__kernel_mode_t` which is aliased to
`unsigned int` rather than `unsigned short` like it is on the other targets.
Update the condition to reflect that.
The SwiftStdint.h header is used in the compiler as well. The compiler may be
built with cl (Visual Studio) on Windows, which does not define
`__INTPTR_TYPE__` nor does it define `__INTPTR_WIDTH__`. Simply define the
`__swift_{,u}intptr_t` typedefs as Microsoft does on that platform when building
with Visual Studio.
If we know a key path component can be accessed as a stored property, then we should also know whether it's a `let` or not, so it should be safe to encode this in the key path pattern. Stage this change in by changing the number of bits used to store in-line offsets, fixing up the parts of the key path implementation that assumed that it took up the entire payload bitfield.
This tells the compiler that repeated calls to this function will always generate the same result, which allows the optimizer to coalesce multiple calls, hoist calls out of loops, and other such nice things.
rdar://problem/20690429
SwiftPrivate/PRNG.swift:
- currently uses `theGlobalMT19937`;
- previously used `arc4random` (see #1939);
- is obsoleted by SE-0202: Random Unification.
This is a bit easier than the fully general case where both the external descriptor and local pattern have captured arguments (because of generics or subscript indices) since we don't have to combine the two argument files in one component.
dispatch_source_create is marked up as non-null but can actually return
null in programming error issues. In order to maintain the nullabilty
contract, these issues will now cause the overlay to assert and crash
rather than return a null pointer out of a non-null call.
Resolves: <rdar://problem/39937177>
Add ICU shims for creating UTexts and performing grapheme breaking
using UTexts. This will allow us to explore multi-encoding grapheme
support throughout string.
- numericValue returns nil instead of .nan for non-numerics
- Remove small-string optimizations from _scalarName that failed on 32-bit archs
- Put case mappings back into U.S.Properties
- Added more sanity tests
Client code can make a best effort at emitting a key path referencing a property with its publicly exposed API, which in the common case will match what the defining module would produce as the canonical key path component representation of the declaration. We can reduce the code size impact of these descriptors by not emitting them when there's no hidden or possibly-resiliently-changed-in-the-past information about a storage declaration, having the property descriptor symbol reference a sentinel value telling client key paths to use their definition of the key path component.
