Repository navigation
Conversation
Stub cuMemcpy*Async used std::memcpy/std::malloc without <cstring>/<cstdlib>, which broke GCC/ASAN/TSAN builds of test_cuda_coro. Also drop the hardcoded developer-machine nvidia/cu13 include so default no-GPU CI compiles against in-header Driver API stubs. Co-authored-by: 李龙飞 <caixuf@users.noreply.github.com>
cursor
Bot
changed the title
caixuf
marked this pull request as ready for review
September 20, 2026 01:03
caixuf
pushed a commit
that referenced
this pull request
Sep 20, 2026
Owner
Author
|
Merged to main as 05a92a6 (local squash push; API merge was flaky). |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
CI run 35051578273 failed on Build & Test (GCC), TSAN, and ASAN while compiling
test_cuda_coro:The compile line also injected
-I/home/caixuf/.local/lib/python3.12/site-packages/nvidia/cu13/includeon GitHub runners (path does not exist). There is noFLOWCORO_ENABLE_CUDAoption;__has_include(<cuda.h>)already chose the in-header stubs, so this-Idid not switch header mode — it only polluted the compile line.Fix
<cstring>and<cstdlib>ininclude/flowcoro/cuda.hso stubcuMemcpy*Async/cuMemAlloc*compile under GCC 13.nvidia/cu13fallback from the header.tests/CMakeLists.txt, add a CUDA include dir only whencuda.hactually exists (FLOWCORO_CUDA_INCLUDE_DIR,CUDA_HOME,CUDA_PATH, or/usr/local/cuda). Default no-GPU CI uses host stubs.No CUDA feature-scope change.
Verify
Local Release (g++-13, no GPU / no toolkit):
FlowCoro: no CUDA toolkit headers found; test_cuda_coro uses host stubstest_cuda_coroCXX_INCLUDES = -I/workspace/include(nonvidia/cu13-I)test_cuda_coroprints[PASS] FlowCoro native CUDA coroutine awaiter passed!ctest --test-dir build --label-exclude network_external: 24/24 passed, includingtest_cuda_coro