This is an alternative to
https://github.com/llvm/llvm-project/pull/122103
In SPIR-V, private global variables have the Private storage class. This
PR adds a new address space which allows frontend to emit variable with
this storage class when targeting this backend.
This is covered in this proposal: llvm/wg-hlsl@4c9e11a
This PR will cause addrspacecast to show up in several cases, like class
member functions or assignment. Those will have to be handled in the
backend later on, particularly to fixup pointer storage classes in some
functions.
Before this change, global variable were emitted with the 'Function'
storage class, which was wrong.
This is an expensive header, only include it where needed. Move some
functions out of line to achieve that.
This reduces time to build clang by ~0.5% in terms of instructions
retired.
Destructor calls were emitted without convergence intrinsics when building for SPIR-V, which means invalid IR since we
mixed controlled and non-controlled convergence.
We can use *Set::insert_range to collapse:
for (auto Elem : Range)
Set.insert(E);
down to:
Set.insert_range(Range);
In some cases, we can further fold that into the set declaration.
When compiling HIP source for AMDGCN flavoured SPIR-V that is expected
to be consumed by the ROCm HIP RT, it's not desirable to set the OpenCL
Kernel CC on `__global__` functions. On one hand, this is not an OpenCL
RT, so it doesn't compose with e.g. OCL specific attributes. On the
other it is a "noisy" CC that carries semantics, and breaks overload
resolution when using [generic dispatchers such as those used by
RAJA](186d4194a5/src/common/HipDataUtils.hpp (L39)).
Adds `@_init_resource_bindings()` function to module initialization that
includes `handle.fromBinding` intrinsic calls for simple resource
declarations. Arrays of resources or resources inside user defined types
are not supported yet.
While this unblocks our progress on [Compile a runnable shader from
clang](https://github.com/llvm/wg-hlsl/issues/7) milestone, this is
probably not the way we would like to handle resource binding
initialization going forward. Ideally, it should be done via the
resource class constructors in order to support dynamic resource binding
or unbounded arrays if resources.
Depends on PRs #110327 and #111203.
Part 1 of #105076
This potentially affects platforms that support comdats other than ELF,
COFF, or wasm, but that is the intention of the FIXME, and if they don't
want this behavior, they probably shouldn't advertise comdat support.
Updates codegen for global destructors and raising exceptions to ensure
that the function pointers being passed are signed using the correct
schema.
Notably this requires that CodeGenFunction::createAtExitStub to return
an opaque Constant* rather than a Function* as the value being emitted
is no longer necessarily a raw function pointer depending on the
configured ABI.
Co-Authored-By: Akira Hatanaka <ahatanaka@apple.com>
Co-Authored-By: John McCall <rjmccall@apple.com>
The `llvm.invariant.start` intrinsic is already overloaded to work with
memory objects in any address space. We simply instantiate the intrinsic
with the appropriate pointer type.
Fixes#94345.
Co-authored-by: Vito Kortbeek <kortbeek@synopsys.com>
This is in effect a revert of f139ae3d93797, as we have since gained a
more sophisticated way of doing extra IRGen with the addition of
RawAddress in #86923.
Summary:
This patch reworks how we handle global constructors in OpenMP.
Previously, we emitted individual kernels that were all registered and
called individually. In order to provide more generic support, this
patch moves all handling of this to the target backend and the runtime
plugin. This has the benefit of supporting the GNU extensions for
constructors an destructors, removing a class of failures related to
shared library destruction order, and allows targets other than OpenMP
to use the same support without needing to change the frontend.
This is primarily done by calling kernels that the backend emits to
iterate a list of ctor / dtor functions. For x64, this is automatic and
we get it for free with the standard `dlopen` handling. For AMDGPU, we
emit `amdgcn.device.init` and `amdgcn.device.fini` functions which
handle everything atuomatically and simply need to be called. For NVPTX,
a patch https://github.com/llvm/llvm-project/pull/71549 provides the
kernels to call, but the runtime needs to set up the array manually by
pulling out all the known constructor / destructor functions.
One concession that this patch requires is the change that for GPU
targets in OpenMP offloading we will use `llvm.global_dtors` instead of
using `atexit`. This is because `atexit` is a separate runtime function
that does not mesh well with the handling we're trying to do here. This
should be equivalent in all cases except for cases where we would need
to destruct manually such as:
```
struct S { ~S() { foo(); } };
void foo() {
static S s;
}
```
However this is broken in many other ways on the GPU, so it is not
regressing any support, simply increasing the scope of what we can
handle.
This changes the handling of ctors / dtors. This patch now outputs a
information message regarding the deprecation if the old format is used.
This will be completely removed in a later release.
Depends on: https://github.com/llvm/llvm-project/pull/71549
This patch converts `ImplicitParamDecl::ImplicitParamKind` into a scoped enum at namespace scope, making it eligible for forward declaring. This is useful for `preferred_type` annotations on bit-fields.
Close https://github.com/llvm/llvm-project/issues/67893
The root cause of the crash is an oversight that we missed the point
that the same module can be imported multiple times. And we should use
`SmallSetVector` instead of `SmallVector` to filter the case.
I just found that we didn't handle the imports in GMF of PMF when we're
generating the init functions for the current module unit. This looks
like a simple oversight and I'm going to fix that in this patch
directly.
Close https://github.com/llvm/llvm-project/issues/56794
And see https://github.com/llvm/llvm-project/issues/67582 for a detailed
backgrond for the issue.
As required by the Itanium ABI, the module units have to generate the
initialization function. However, the importers are allowed to elide the
call to the initialization function if they are sure the initialization
function doesn't do anything.
This patch implemented this semantics.
Turned out we were making overly simple assumptions about which sections (& section flags) would be used when emitting a global into a custom section. This lead to sections with read-only flags being used for globals of struct types with mutable members.
Fixed by porting the codegen function with the more nuanced handling/checking for mutable members out of codegen for use in the sema code that does this initial checking/mapping to section flags.
Differential Revision: https://reviews.llvm.org/D156726
Partial progress towards replacing in-tree uses of `Type::getPointerTo()`.
This needs to be done before deprecating the API.
Reviewed By: nikic, barannikov88
Differential Revision: https://reviews.llvm.org/D152321
The original name "ASTContext::getNamedModuleForCodeGen" is not properly
reflecting the usage of the interface. This interface can be used to
judge the current module unit in both sema analysis and code generation.
So the original name was not so correct.
Decl::isInCurrentModuleUnit
Refactor `Sema::isModuleUnitOfCurrentTU` to `Decl::isInCurrentModuleUnit`
to make code simpler a little bit. Note that although this patch
introduces a FIXME, this is an existing issue and this patch just tries
to describe it explicitly.
We need to be able to distinguish individual TUs from the same module in cases
where TU-local entities either need to be hidden (or, for some cases of ADL in
template instantiation, need to be detected as exposures).
This creates a module type for the implementation which implicitly imports its
primary module interface per C++20:
[module.unit/8] 'A module-declaration that contains neither an export-keyword
nor a module-partition implicitly imports the primary module interface unit of
the module as if by a module-import-declaration.
Implementation modules are never serialized (-emit-module-interface for an
implementation unit is diagnosed and rejected).
Differential Revision: https://reviews.llvm.org/D126959
This reverts commit c6e9823724ef6bdfee262289ee34d162db436af0.
Reason: Broke the ASan buildbots, see https://reviews.llvm.org/D126959
(the original phabricator review) for more info.
We need to be able to distinguish individual TUs from the same module in cases
where TU-local entities either need to be hidden (or, for some cases of ADL in
template instantiation, need to be detected as exposures).
This creates a module type for the implementation which implicitly imports its
primary module interface per C++20:
[module.unit/8] 'A module-declaration that contains neither an export-keyword
nor a module-partition implicitly imports the primary module interface unit of
the module as if by a module-import-declaration.
Implementation modules are never serialized (-emit-module-interface for an
implementation unit is diagnosed and rejected).
Differential Revision: https://reviews.llvm.org/D126959
This follows 2b4fa53 which made Clang not emit destructor calls for such
objects. However, they would still not get emitted as constants since
CodeGenModule::isTypeConstant() returns false if the destructor is
constexpr. This change adds a param to make isTypeConstant() ignore the
dtor, allowing the caller to check it instead.
Fixes Issue #61212
Differential revision: https://reviews.llvm.org/D145369
Previously we'll set the named modules for ASTContext in ParseAST. But
this is not intuitive and we need comments to tell the intuition. This
patch moves the code the right the place, where the corrresponding
module is first created/loaded. Now it is more intuitive and we can use
the value in the earlier places.
For the Itanium ABI, we emit an initializer for each module. This is responsible
for handling initialization of global vars. Relates to P1874R1.
The initializer has a known mangling and is automatically called from any TU that
imports a module. Since, at present, the importer has no way to determine that an
imported module does not require an initializer, we generate the initializer for
all cases (even when it is empty).
Initializers must be run once, with the ordering guaranteed by the import graph
and this is ensured in the current code by addition of a guard variable.
In the case that a module has no requirement for global initializers, and also does
not import any other modules, we can elide the guard variable.
Differential Revision: https://reviews.llvm.org/D134589
This value was added to clang/Basic in D111566, but is only used during
codegen, where we can use the LLVM IR DataLayout instead. I noticed this
because the downstream CHERI targets would have to also set this value
for AArch64/RISC-V/MIPS. Instead of duplicating more information between
LLVM IR and Clang, this patch moves getTargetAddressSpace(QualType T) to
CodeGenTypes, where we can consult the DataLayout.
Reviewed By: rjmccall
Differential Revision: https://reviews.llvm.org/D138296
This revision fixes typos where there are 2 consecutive words which are
duplicated. There should be no code changes in this revision (only
changes to comments and docs). Do let me know if there are any
undesirable changes in this revision. Thanks.
Fixes https://github.com/llvm/llvm-project/issues/55804
The lexing order is already bookkept in DelayedCXXInitPosition but we
were not using it based on the wrong assumption that inline variable is
unordered. This patch fixes it by ordering entries in llvm.global_ctors
by orders in DelayedCXXInitPosition.
for llvm.global_ctors entries without a lexing order, ordering them by
the insertion order.
(This *mostly* orders the template instantiation in
https://reviews.llvm.org/D126341 intuitively, minus one tweak for which I'll
submit a separate patch.)
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D127233
Currently we treat initializers with init_seg(compiler/lib) as similar
to any other init_seg, they simply have a global variable in the proper
section (".CRT$XCC" for compiler/".CRT$XCL" for lib) and are added to
llvm.used. However, this doesn't match with how LLVM sees normal (or
init_seg(user)) initializers via llvm.global_ctors. This
causes issues like incorrect init_seg(compiler) vs init_seg(user)
ordering due to GlobalOpt evaluating constructors, and the
ability to remove init_seg(compiler/lib) initializers at all.
Currently we use 'A' for priorities less than 200. Use 200 for
init_seg(compiler) (".CRT$XCC") and 400 for init_seg(lib) (".CRT$XCL"),
which do not append the priority to the section name. Priorities
between 200 and 400 use ".CRT$XCC${Priority}". This allows for
some wiggle room for people/future extensions that want to add
initializers between compiler and lib.
Fixes#56922
Reviewed By: rnk
Differential Revision: https://reviews.llvm.org/D131910
Previously when we add module initializer, we forget to handle header
units. This results that we couldn't compile a Hello World Example with
Header Units. This patch tries to fix this.
Reviewed By: iains
Differential Revision: https://reviews.llvm.org/D130871
This reverts commit db6152ad66d7cf48f9f5c3eb28bf54c092978773.
This commit fails in ppc64. Since we want to backport it to 15.x. So
revert it now to keep the patch complete.
