Move `CallConv` from `CUDAContext` to `FunctionDescriptor` #717

isVoid · 2026-01-13T17:55:11Z

Today, calling conventions are defined globally per compilation context. This makes it hard to switch flexibly between the Numba ABI and the C ABI when declaring external functions. It also explains the need for the kernel “fixup” logic: CUDA kernels are fundamentally C-ABI, but have historically been forced through the Numba ABI path.

This PR moves calling-convention selection to a more granular level, removing these limitations and eliminating the kernel fixup workaround. It also lays the groundwork for users to plug in additional calling-convention implementations in the future.

copy-pr-bot · 2026-01-13T17:55:15Z

Auto-sync is disabled for ready for review pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

greptile-apps · 2026-01-13T17:59:25Z

Greptile Overview

Greptile Summary

This PR successfully refactors the calling convention architecture by moving CallConv from CUDAContext to FunctionDescriptor, enabling per-function calling convention specification. This architectural improvement allows different functions to use different ABIs (Numba vs C) and eliminates the need for kernel "fixup" logic.

Key Changes:

Moved CUDACallConv and CUDACABICallConv classes from target.py to callconv.py
Added call_conv and abi_info fields to FunctionDescriptor and CUDAFlags for per-function calling convention storage
Changed CUDATargetContext.call_conv from cached property to property reading from fndesc.call_conv
Updated declare_device to accept abi parameter ("numba" or "c")
Removed cabi_wrap_function as C ABI functions now work directly without wrapping
Added comprehensive tests for C ABI device functions with 0, 1, 2 arguments and void return

Critical Issue Found:

Bug in imputils.py line 199: The fix_returning_optional function accesses status.is_none without checking if status is None first. When using C ABI calling convention, CUDACABICallConv.call_function returns status = None (line 417 in callconv.py), which will cause an AttributeError when the code tries to access status.is_none.

Confidence Score: 2/5

This PR contains a critical bug that will cause runtime errors when using C ABI functions with optional return types
The architectural refactoring is well-designed and most changes are clean, but there is a critical bug in imputils.py line 199 that will cause AttributeError when calling C ABI device functions that have optional return types. The fix_returning_optional function assumes status is always an object with an is_none attribute, but for C ABI functions, status is None.
numba_cuda/numba/cuda/core/imputils.py requires immediate attention - the bug on line 199 must be fixed before merge

Important Files Changed

File Analysis

Filename	Score	Overview
numba_cuda/numba/cuda/core/imputils.py	2/5	Changed to use `fndesc.call_conv` instead of `context.call_conv` and added null check for status. Contains critical bug: `fix_returning_optional` accesses `status.is_none` without checking if status is None first (line 199).
numba_cuda/numba/cuda/core/callconv.py	5/5	Added `CUDACallConv` and `CUDACABICallConv` classes moved from `target.py`. Clean implementation of two calling conventions with proper separation of concerns.
numba_cuda/numba/cuda/core/funcdesc.py	5/5	Added `call_conv` and `abi_info` fields to `FunctionDescriptor` for per-function calling convention storage. Changes enable ABI name handling via `abi_info`.
numba_cuda/numba/cuda/target.py	5/5	Removed `CUDACallConv` and `CUDACABICallConv` classes (moved to `callconv.py`). Changed `call_conv` from cached property to property reading from `fndesc.call_conv`.
numba_cuda/numba/cuda/compiler.py	4/5	Removed `cabi_wrap_function` (no longer needed). Updated `compile_cuda` to accept `abi` parameter and set calling convention in flags. Updated `declare_device_function` to instantiate appropriate calling convention based on ABI.
numba_cuda/numba/cuda/core/compiler.py	4/5	Added import for `CUDACallConv` and updated error model creation to pass a `CUDACallConv` instance instead of context.

greptile-apps

_{4 files reviewed, 4 comments}

_{Edit Code Review Agent Settings | Greptile}

numba_cuda/numba/cuda/tests/cudadrv/test_reset_device.py

greptile-apps · 2026-01-13T17:59:33Z

numba_cuda/numba/cuda/tests/cudapy/test_multithreads.py


 @skip_under_cuda_memcheck("Hangs cuda-memcheck")
 @skip_on_cudasim("disabled for cudasim")
+@unittest.skip


Test skipped without explanation - check if the refactoring broke this test

Some of the multithread / multigpu test were suspected to be not tested properly in the current CI system as they are all single-GPU runners. They show up as failures in my local environment but was not related this PR.

greptile-apps · 2026-01-13T17:59:34Z

numba_cuda/numba/cuda/compiler.py

    )
    resty = cres.signature.return_type

    if resty and not device and resty != types.void:
        raise TypeError("CUDA kernel must have void return type.")


Removed cabi_wrap_function call for device functions with C ABI. Verify this was intentional and that device functions with abi="c" still work correctly without the wrapper.

This is indeed intentional, because function / kernels can bare its own calling convention, rather than using a target-wide calling convention. cabi_wrap_function was meant to patch a function with Numba ABI to a caller that expects a CABI function. The fix in this PR is imply switching the kernel to bare a CUDACABICallConv and should work as expected.

greptile-apps · 2026-01-13T17:59:37Z

Additional Comments (1)

numba_cuda/numba/cuda/core/imputils.py
Bug: status can be None when using C ABI calling convention (see line 222 check and CUDACABICallConv.call_function which returns status = None), but this line tries to access status.is_none, which will raise AttributeError.

        if status is not None and builder.not_(status.is_none):

greptile-apps

_{1 file reviewed, 1 comment}

_{Edit Code Review Agent Settings | Greptile}

greptile-apps · 2026-01-13T18:13:12Z

Additional Comments (1)

numba_cuda/numba/cuda/core/imputils.py
status can be None when using C ABI (see line 222 check and CUDACABICallConv.call_function returns status = None), causing AttributeError when accessing status.is_none

        if status is not None and builder.not_(status.is_none):

greptile-apps

_{1 file reviewed, 1 comment}

_{Edit Code Review Agent Settings | Greptile}

greptile-apps · 2026-01-13T18:17:52Z

Additional Comments (1)

numba_cuda/numba/cuda/core/imputils.py
status can be None when using C ABI calling convention (see line 222 check and CUDACABICallConv.call_function which returns status = None in callconv.py line 417), causing AttributeError when accessing status.is_none

        if status is not None and builder.not_(status.is_none):

gmarkall · 2026-01-15T13:08:07Z

/ok to test

isVoid added 4 commits December 18, 2025 09:59

checkpointing 121825

bcee996

checkpointing 010626

a988ca3

initial

1419998

Merge remote-tracking branch 'origin' into experimental-callconv

250886f

greptile-apps bot reviewed Jan 13, 2026

View reviewed changes

remove stale skip

e60bf0f

greptile-apps bot reviewed Jan 13, 2026

View reviewed changes

remove cabi_wrap_function

72f8f8e

isVoid changed the title ~~Move CallConv from CUDAContext to FunctionDescriptor~~ Move CallConv from CUDAContext to FunctionDescriptor Jan 13, 2026

greptile-apps bot reviewed Jan 13, 2026

View reviewed changes

gmarkall added the 3 - Ready for Review Ready for review by team label Jan 15, 2026

Move CallConv from CUDAContext to FunctionDescriptor #717

Are you sure you want to change the base?

Move CallConv from CUDAContext to FunctionDescriptor #717

Uh oh!

Conversation

isVoid commented Jan 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

copy-pr-bot bot commented Jan 13, 2026

Uh oh!

greptile-apps bot commented Jan 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Greptile Overview

Greptile Summary

Confidence Score: 2/5

Important Files Changed

Uh oh!

greptile-apps bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

greptile-apps bot Jan 13, 2026

Choose a reason for hiding this comment

Uh oh!

isVoid Jan 13, 2026

Choose a reason for hiding this comment

Uh oh!

greptile-apps bot Jan 13, 2026

Choose a reason for hiding this comment

Uh oh!

isVoid Jan 13, 2026

Choose a reason for hiding this comment

Uh oh!

greptile-apps bot commented Jan 13, 2026

Uh oh!

greptile-apps bot left a comment

Choose a reason for hiding this comment

Uh oh!

greptile-apps bot commented Jan 13, 2026

Uh oh!

greptile-apps bot left a comment

Choose a reason for hiding this comment

Uh oh!

greptile-apps bot commented Jan 13, 2026

Uh oh!

gmarkall commented Jan 15, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Move `CallConv` from `CUDAContext` to `FunctionDescriptor` #717

Move `CallConv` from `CUDAContext` to `FunctionDescriptor` #717

isVoid commented Jan 13, 2026 •

edited

Loading

greptile-apps bot commented Jan 13, 2026 •

edited

Loading