This updates the CIR constant emitter to use the correct destination
type when emitting a constant initializer for a structure that might be
initialized with non-prototyped function pointers. We were previously
using the type from whatever function declaration we had, but this may
not be the correct type.
This change also updates the `replaceUsesOfNonProtoTypeWithRealFunction`
to ignore global initializer uses, which do not need to be updated after
this change.
This is a pretty simple one, its just a type of decl-context. The actual
exporty-ness is handled on a per-declaration basis.
This patch just makes sure we emit them, as I suspect this will reveal
quite a bit more issues in module code I suspect.
Upstreaming clangIR PR: https://github.com/llvm/clangir/pull/422
This PR implement support for `__attribute__((section("name")))` on
global variables in ClangIR, matching OGCG behavior.
This patch adds OpenMP attributes to the MLIR module op used to
communicate various settings e.g. thread options and offloading
information.
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
We had an errorNYI diagnostic to trigger when we generated an alias for
a ctor or dtor that had an existing declaration. Because functions are
used via flat symbol references, all that is needed is to erase the old
declaration. This change does that.
Upstreaming clangIR PR: https://github.com/llvm/clangir/pull/2091
This patch adds support for AMDGPU-specific function attributes for HIP
kernels
Added setTargetAttributes for AMDGPUTargetCIRGenInfo to set kernel
attributes
Added generic string attribute handler in amendFunction to translate
string-values with "cir." prefix function attributes to LLVM function
attributes
Follows OGCG AMDGPU implementation from
"clang/lib/CodeGen/Targets/AMDGPU.cpp".
getNaturalTypeAlignment on a reference type returned pointer alignment
instead of pointee alignment. Pass the pointee type with
forPointeeType=true to match traditional codegen's
getNaturalPointeeTypeAlignment behavior. Fix applies to both argument
and return type attribute construction paths.
Add the amendOperation override to handle CIR dialect attributes during
MLIR-to-LLVM IR translation. This dispatches to amendModule for ModuleOp,
enabling module metadata.
This PR also adds support to emit AMDGPU-specific module flags
amdhsa_code_object_version and amdgpu_printf_kind to match OGCG
behavior.
In CIRGenModule, the flags are stored as CIR module attributes:
cir.amdhsa_code_object_version (integer)
cir.amdgpu_printf_kind (string: "hostcall" or "buffered")
During lowering to LLVM IR (in LowerToLLVMIR.cpp), these attributes are
converted to LLVM module flags.
Upstreaming basic changes from clangIR PRs:
61e9ebd9f8https://github.com/llvm/clangir/pull/768https://github.com/llvm/clangir/pull/773https://github.com/llvm/clangir/pull/2100
This change upstreams the CIR implementation of global variable
replacement handling for cases where the global was used in a
cir.global_view operation, either as an initializer for another global
or as a constant ptr.
In preperation of actually lowering data members as fields to a record
type, this patch does a minor refactor to make their single current use
have a slightly simpler interface. This will prevent us from having to
copy/paste this later.
Also, this patch removes a pair of now-orphaned builders, instead
preferring to use the ones that come from the parent builder type.
This adds support for setting the `nounwind` attribute on declarations
and call sites in CIR.
Currently, we have both `nothrow` and `nounwind` in CIR. I've chosen to
use `nothrow` in this PR because it was the most common. I plan to
combine them, using `nounwind` everywhere since that's the LLVM IR
spelling, but that's a more invasive so I'd like to defer it to a future
change.
This adds code to set the `builtin` attribute when needed on operator
new and delete calls. This also required setting `nobuiltin` for
replaceable global allocation functions and threading the `builtin`
attribute through the LLVM dialect.
This is used somewhat rarely, but is a pretty simple emission of
pointers, and ends up using infrastructure we already have.
Additionally, this is the first use of `getNaturalTypeAlignment` that
uses the `pointee` argument, so this adds the implementation there,
which includes some alignment work for CXXRecordDecls, so this
implements that as well.
We are currently only emitting Vtables that have an 'immediate' need to
emit. There rest, we are supposed to add to a list and emit at the end
of the translation unit if necessary. This patch implements that
infrastructure.
The test added is from classic-codegen and came in at the same time as
the deferred vtable emission over there, and only works with deferred
vtable emission, and while it does test the deferred emission, tests
quite a bit more than that. AND since it came in with the same
functionality in classic codegen, seemed to make sense to come in here
too.
Our current implementation of string lowering did some work to remove
extra trailing zeros, plus do a 'zero' constant. That is unchanged by
this patch. However, this patch ALSO ensures that we do the 'remove
extra trailing zeros' to remove ALL trailing zeros, which likely has
canonicalization benefits later on.
However, the real benefit of this patch is to make string emission by
default emit a null-terminator, which fixes the virtual table 'name'
field get lowered correctly. We do this by making the builder::getString
function take an argument (true by default) that will ensure we add a
null terminator if necessary.
This reflects the llvm::ConstantDataArray::getString function, which has
the same functionality. However, doing this during lowering seems
incorrect, since the FE is the one that knows whether these null
terminators are necessary. There is not currently an 'opt out' use of
the behavior, but the functionality is left in place to better reflect
the ConstantDataArray::getString function interface.
Note with the tests that this fixes some inconsistencies between LLVM
and OGCG lowering.
This patch adds typeid lowering, which uses a lot of the infrastructure
from dynamic_cast. However, this adds a `get_type_info` operation that
gets the type info out of a vtable pointer as well, which lets the
offset be handled by the ABI specific lowering code.
This change upstreams the CIR implementation of global variable
replacement. When we get a call to get or create a global variable using
a type that does not match the previous type for a variable of the same
name, we need to replace the old definition with the new one. In classic
codegen that was as simple as replaceAllUses+eraseFromParent, but in CIR
because we have typed pointers, we need to visit the uses and update
them with bitcasts to reflect the new type.
This implements vtable thunk handling in CIR based on the incubator
code, but also compared against the latest Clang LLVM IR codegen.
Eventually, we'll want to create CIR abstractions for all of this and
move the CXXABI-specific details into the CXXABI lowering pass. For now,
we just implement it directly in codegen.
These are fairly simple, particularly if they don't need special
cleanups (which is left unimplemented), but this provides init for a
global reference variable.
Similar to what I did for return types in #181052, this patch adds
support for 4 of the function attributes on arguments (noundef, nonnull,
dereferenceable, align). The logic for these is all pretty similar
(though SLIGHTLY different enough from eachother unfortunately), so they
are being submitted together. This handles 'this' and normal arguments.
This patch implements the infrastructure for return attributes on
function/call operations, a little of the common infrastructure for arg
attributes on the same, and 4 return attributes lowering: noundef
nonnull
dereferenceable
align
These 4 common attributes are all pretty reasonable/common, so these
will change a lot of tests.
This patch chooses to just use the LLVM-IR-Dialect variant of these
attributes (as a NamedAttr), which means no changes to the dialect or
lowering are necessary.
This implements the lowering of static local variables with the Itanium C++ ABI
guard variable pattern in LoweringPrepare. This is initial support, errorNYI covering all that hasn't been added just yet.
When a GlobalOp has the static_local attribute and a ctor region, this pass:
1. Creates a guard variable global (mangled name from AST)
2. Inserts the guard check pattern at each GetGlobalOp use site:
- Load guard byte with acquire ordering
- If zero, call __cxa_guard_acquire
- If acquire returns non-zero, inline the ctor region code
- Call __cxa_guard_release
3. Clears the static_local attribute and ctor region from the GlobalOp
Once the new design doc lands I'll add more information over there.
This adds CIRGen infrastructure for C++ function-local static variables
that require guarded initialization (Itanium C++ ABI).
Changes:
- Add ASTVarDeclAttr to carry VarDecl AST through the pipeline
- Add emitGuardedInit() to CIRGenCXXABI for guarded initialization
- Add emitCXXGuardedInit() to CIRGenFunction
- Replace NYI in addInitializerToStaticVarDecl() with ctor region emission
- Set static_local attribute on GlobalOp and GetGlobalOp
This doesn't unwrap the high level constructs just yet - The global's ctor region contains the initialization code, which will be lowered by LoweringPrepare to emit the actual guard variable pattern with __cxa_guard_acquire/__cxa_guard_release calls.
related: #175871
This patch adds foundational infra for device-side CUDA/HIP compilation
by introducing NVPTX target info and implementing the global emission
filtering logic.
NVPTX Target Info to allows us to compile against that triple:
- Add NVPTXABIInfo and NVPTXTargetCIRGenInfo classes
- Wire up nvptx and nvptx64 triples in getTargetCIRGenInfo()
- Add createNVPTXTargetCIRGenInfo() factory function
CUDA/HIP Global Emission Filtering (most of this is boilerplate from the
AST) This basically narrows down to:
- Skip host-only functions (no `__device__` attribute) when
`-fcuda-is-device`
- Skip device-only functions (device without host) on host side
- Always emit ` __global__` kernels and `__host__` `__device__`
functions on both sides
- Add `shouldEmitCUDAGlobalVar()` to handle variable emission
(device/constant/shared variables)
- Handle special cases: implicit host/device templates, lambda call
operators
Part of #175871
This patch adds the initial implementation of the CUDA/NV Runtimes
generating code for the device stub body. tested on CUDA. HIP coverage
to be added in a later PR.
Previous efforts in both the incubator and upstreamed added the 'pure'
and 'const' attribute lowering to calls. This patch completes that
effort by making it also appear on functions with this attribute.
While looking into the work required to add attributes to
functions/arguments/etc, I discovered a lot of overlap with calling
conventions, which I know others are working on. Rather than risk a
later conflict, this patch adds some basic infrastructure that will
hopefully limit the fallout. This patch is NFC as it doesn't modify the
IR at all, just adds a few calls to the 'right place' and adds some
arguments.
Additionally, it adds the calling convention type to the IR, but leaves
it empty, as it is useful to be able to pass around but not particularly
possible to define/make do anything yet.
Part of: #153267
This is a continuation of: #154014
This patch handles output operands for inline assembly, taking the
original patch, but adding some additional tests, plus responding to all
of the original comments, plus doing some of my personal cleanup
(including extracting addVariableConstraints in a separate patch).
---------
Co-authored-by: Iris Shi <0.0@owo.li>
Implement emitDeclInvariant to emit llvm.invariant.start intrinsic for
global variables with constant storage. This enables optimizations by
marking when a global becomes read-only after initialization.
## Changes
- Add emitDeclInvariant and emitInvariantStart functions in
CIRGenCXX.cpp
- Add emitInvariantStart declaration in CIRGenFunction.h
- Update emitCXXGlobalVarDeclInit to call emitDeclInvariant for constant
storage globals after initialization
- Update getOrCreateCIRGlobal to set constant flag on globals with
constant storage
- Add comprehensive test covering positive and negative cases
## Implementation Details
The implementation handles address spaces correctly, dynamically
constructing the intrinsic name (e.g., invariant.start.p0,
invariant.start.p10) based on the pointer's address space. The intrinsic
is only emitted when optimizations are enabled, matching classic codegen
behavior.
## Testing
All tests pass (411/412, 1 unsupported). The test file includes CIR,
LLVM, and OGCG checks for both optimized and non-optimized builds.
This patch adds the basic infrastructure for lowering an OpenMP
directive, which should enable someone to take over the OpenMP lowering
in the future. It adds the lowering entry points to CIR in the same way
as OpenACC.
Note that this does nothing with any of the directives, which will
happen in a followup patch. No infrastructure for clauses is added
either, but that will come in a followup patch as well.
This PR migrates support for wide string literals from the incubator to
upstream.
## Changes
- Implement wide string literal support in
`getConstantArrayFromStringLiteral`
- Handle wchar_t, char16_t, and char32_t string literals
- Collect code units and create constant arrays with IntAttr elements
- Use ZeroAttr for null-filled strings
## Testing
- Copied `wide-string.cpp` test file from incubator
- Expanded test to include wchar_t test cases (incubator only had
char16_t and char32_t)
- All tests pass
---------
Co-authored-by: Andy Kaylor <akaylor@nvidia.com>
The bind clause specifies the name of the function to call on the
device, and takes either a string or identifier(per the standard):
"If the name is specified as an identifier, it is callled as if the name
were specified in the language being compiled. If the name is specified
as a string, the string is used for the procedure name unmodified".
The latter (as a string) is already implemented, this patch implements
the former. Unfortunately, no existing implementation of this in C++
seems to exist. Other languages, the 'name' of a function is sufficient
to identify it (in this case 'bind' can refer to undeclared functions),
so it is possible to figure out what the name should be. In C++ with
overloading (without a discriminator, ala-fortran), a name only names an
infinite overload set.
SO, in order to implement this, I've decided that the 'called as'
(bound) function must have the same signature as the one marked by the
'routine'. This is trivially sensible in non-member functions, however
requires a bit more thought for member(and thus lambda-call-operators)
functions. In this case, we 'promote' the type of the function to a
'free' function by turning the implicit 'this' to an explicit 'this'.
I believe this is the most sensible and reasonable way to implement
this, and really the only way to make something usable.
This commit adds full support for thread-local storage variables in
ClangIR, including code generation, lowering to LLVM IR, and
comprehensive testing.
Changes include:
- Added CIR_TLSModel enum with 4 TLS models (GeneralDynamic,
LocalDynamic, InitialExec, LocalExec) to CIROps.td
- Extended GlobalOp with optional tls_model attribute
- Extended GetGlobalOp with thread_local unit attribute
- Added verification to ensure thread_local GetGlobalOp references
globals with tls_model set
- Implemented GetDefaultCIRTLSModel() and setTLSMode() in CIRGenModule
- Updated getAddrOfGlobalVar() to handle TLS access
- Removed MissingFeatures assertions for TLS operations
- Added lowering of GetGlobalOp with TLS to llvm.threadlocal.address
intrinsic
- Added lowering of GlobalOp with tls_model to LLVM thread_local globals
- Added comprehensive test with CIR, LLVM, and OGCG checks
Known limitations (matching incubator):
- Static local TLS variables not yet implemented
- TLS_Dynamic with wrapper functions not yet implemented
Fixes#153270
This PR upstreams support for the `cir.indirectBr` operation, which is
used to implement GCC’s labels-as-values `indirect goto`.
To ensure correct lowering, we introduce precise bookkeeping to
associate each `block_address` operation with its corresponding `label`
op. This is required because a `block_address` may be emitted before the
`label` it refers to. In such cases, the reference is deferred and later
resolved by `resolveBlockAddresses`, which guarantees that all
`indirectBr` successors are emitted in the correct and fully resolved
order.
The 'routine' construct just adds a acc.routine element to the global
module, which contains all of the information about the directive. it
contains a reference to the function, which also contains a reference to
the acc.routine, which this generates.
This handles both the implicit-func version (where the routine is
spelled without parens, and just applies to the next function) and
the explicit-func version (where the routine is spelled with the func
name in parens).
The AST stores the directive in an OpenACCRoutineDeclAttr in the
implicit case, so we can emit that when we hit the function declaration.
The explicit case is held in an OpenACCRoutineAnnotAttr on the function,
however, when we emit the function we haven't necessarily seen the
construct yet, so we can't depend on that attribute. Instead, we save up
the list in Sema so that we can emit them all at the end.
This results in the tests getting really hard to read (because ordering
is a little awkward based on spelling, with no way to fix it), so we
instead split the tests up based on topic.
One last thing: Flang spends some time determining if the clause lists
of two routines on the same function are identical, and omits the
duplicates. However, it seems to do a poor job on this when the ordering
isn't the same, or references are slightly different. This patch doesn't
bother trying that, and instead emits all, trusting the ACC dialect to
remove duplicates/handle duplicates gracefully.
Note; This doesn't cause emission of functions that would otherwise not
be emitted, but DOES emit routine references based on which function
they are attached to.
