Extended existential type shapes can trigger this by introducing
more entities (and thus causing GlobalVars to be rehashed) during
the lazy-emission callback.
Fixes rdar://98995607
Include the parent `ModuleDecl` when serializing a `SILFunction` so that it is available on deserialized functions even though the full `DeclContext` is not present. With the parent module always available we can reliably compute whether the `SILFunction` comes from a module that was imported `@_weakLinked`.
Serialize the `DeclContext` member of `SILFunction` so that it can be used to look up the module that a function belongs to in order to compute weak import status.
Resolves rdar://98521248
Trivial conflict caused by the line above the
`IGM.constructInitialFnAttributes` change in `lib/IRGen/GenDecl.cpp`
having an extra argument passed in rebranch (due to the new LLVM API).
Adds frontend option -enable-stack-protector to enable emission of a
stack protector.
Disabled by default.
When enabled enables LLVM's strong stack protection mode.
rdar://93677524
Put pointers to class_ro_t referenced from generic class patterns in a section __swift_rodatas such that they are discoverable by the linker.
The linker can then make the method lists contained in the class_ro_t relative like it can for objective c class metadata from non-generic swift classes.
rdar://66634459
LLVM's O0 pipeline has decided to run GlobalDCE (if there are coro intrinsics present in the module).
Add definitions to compiler.used to counter act that.
So far, static arrays had to be put into a writable section, because the isa pointer and the (immortal) ref count field were initialized dynamically at the first use of such an array.
But with a new runtime library, which exports the symbols for the (immortal) ref count field and the isa pointer, it's possible to put the whole array into a read-only section. I.e. make it a constant global.
rdar://94185998
This reverts the revert commit df353ff3c0.
Also, I added a frontend option to disable this optimization: `-disable-readonly-static-objects`
So far, static arrays had to be put into a writable section, because the isa pointer and the (immortal) ref count field were initialized dynamically at the first use of such an array.
But with a new runtime library, which exports the symbols for the (immortal) ref count field and the isa pointer, it's possible to put the whole array into a read-only section. I.e. make it a constant global.
rdar://94185998
The `AttrBuilder` API was changed in llvm/llvm-project
d2cc6c2d0c2f8a6e272110416a3fd579ed5a3ac1 to now take a `LLVMContext &`
as part of its constructor. Update uses to pass one in.
`PointerType::getElementType` has been removed entirely as part of the
opaque pointers migration. Update to `getPointerElementType` for now
until we've also migrated.
Enable testing makes `internal` types visible from outside the module.
We can no longer treat super classes as resilient.
Follow-up to #41044.
rdar://90489618
On some Harvard architectures like WebAssembly that allow sliding code
and data address space offsets independently, it's impossible to make
direct relative reference to code from data because the relative offset
between them is not representable.
Use absolute function references instead of relative ones on such targets.
Since marker protocol do not exist at runtime, it's harmful to keep them
in a generic environment because they cannot be checked.
Resolves: rdar://90293494
The main point of this change is to make sure that a shared function always has a body: both, in the optimizer pipeline and in the swiftmodule file.
This is important because the compiler always needs to emit code for a shared function. Shared functions cannot be referenced from outside the module.
In several corner cases we missed to maintain this invariant which resulted in unresolved-symbol linker errors.
As side-effect of this change we can drop the shared_external SIL linkage and the IsSerializable flag, which simplifies the serialization and linkage concept.
The linkage inferred by internal linking is overly restrictive. It uses
the "internal linkage type" which is for translation unit local symbols,
that is `static` storage. These symbols participate in linking and need
to be given external storage but hidden visibility and no DLL storage in
static linking. This permits them to participate in symbolic resolution
during linking but are made module local, not LLVM module local.
This improves the static linking story on Windows by wiring up
information about the static vs dynamic build of the module into the
code generation for references to module external symbols. We now can
indicate a module external symbol to be known local if it is expected to
be internalized due to static linking. Even though the symbol may not
be defined in the same module, the definition will be provided locally
at link time. This avoids needing to resolve a dllimport'ed symbol to a
local symbol at link time.
This pipes the `-static` flag when building a static library into IRGen.
This should have no impact on non-Windows targets as the usage of the
information simply removes the `dllexport` attribute on the generated
interfaces. This ensures that a library built with `-static` will not
re-export its interfaces from the consumer. This is important to ensure
that the consumer does not vend the API surface when it statically links
a library. In conjunction with the removal of the force load symbol,
this allows the generation of static libraries which may be linked
against on Windows. However, a subsequent change is needed to ensure
that the consumer does not mark the symbol as being imported from a
foreign module (i.e. `dllimport`).
The special convention currently means three things:
1. There is no async FP symbol for the function. Calls should go directly
to the function symbol, and an async context of fixed static size should be
allocated. This is mandatory for calling runtime-provided async functions.
2. The callee context should be allocated but not initialized. The main
context pointer passed should be the caller's context, and the continuation
function pointer and callee context should be passed as separate arguments.
The function will resume the continuation function pointer with the caller's
context. This is a micro-optimization appropriate for functions that are
expected to frequently return immediately; other functions shouldn't bother.
3. Generic arguments should be suppressed. This is a microoptimization for
certain specific runtime functions where we happen to know that the runtime
already stores the appropriate information internally. Other functions
probably don't want this.
Obviously, these different treatments should be split into different
predicates so that functions can opt in to different subsets of them.
I've also set the code up so that runtime functions can more easily
request a specific static async context size. Previously, it was a
confusingly embedded assumption that the static context size was always
exactly two pointers more than the header.
The right index is the index derived for calling a function
(Signature::forAsyncEntry) rather then for receiving the results
(Signature::forAsyncAwait(); the resume fn signature).
rdar://86863272
Direct pointer to the accessor cannot be called at runtime,
so here is how everything is stored:
- For `distributed` and `async` functions -> async function pointer;
- For regular functions -> function pointer.
