Maintenance: small CI and test-helper cleanups (#2126)

* ci: add 60 minute timeouts to test jobs

Add explicit 60 minute GitHub Actions timeouts to the reusable test workflows so hung test runs release runners promptly. This intentionally covers only test jobs because build jobs are expected to be extremely unlikely to hang, while the recent CI failures came from tests wedging for hours.

* fix trivial spelling errors in two files

Fix trivial spelling errors, including errors pointed out in PR #2103.

* Align WSL detection helpers

Remove the environment-variable fallback from cuda_python_test_helpers._detect_wsl so the test helper matches cuda.core's /proc-based WSL detection. Add comments to both copies to keep the duplicated implementations synchronized.

Xref: https://github.com/NVIDIA/cuda-python/pull/2118/changes#r3278720224

Author: rwgk

.github/workflows/test-sdist-linux.yml

@@ -22,6 +22,7 @@ permissions:
 jobs:
   test-sdist:
     name: Test sdist builds
+    timeout-minutes: 60
     runs-on: linux-amd64-cpu8
     steps:
       - name: Checkout ${{ github.event.repository.name }}

.github/workflows/test-sdist-windows.yml

@@ -28,6 +28,7 @@ permissions:
 jobs:
   test-sdist:
     name: Test sdist builds
+    timeout-minutes: 60
     runs-on: windows-2022
     steps:
       - name: Checkout ${{ github.event.repository.name }}

.github/workflows/test-wheel-linux.yml

@@ -96,6 +96,7 @@ jobs:
 
   test:
     name: Python ${{ matrix.PY_VER }}, CUDA ${{ matrix.CUDA_VER }} (${{ (matrix.LOCAL_CTK == '1' && 'local') || 'wheels' }}), GPU ${{ matrix.GPU }}${{ matrix.GPU_COUNT != '1' && format(' (x{0})', matrix.GPU_COUNT) || '' }}${{ matrix.FLAVOR && format(', {0}', matrix.FLAVOR) || '' }}${{ matrix.TORCH_VER && format(', {0}', matrix.TORCH_VER) || '' }}${{ matrix.MODE == 'nightly-numba-cuda' && ', latest' || '' }}
+    timeout-minutes: 60
     needs: compute-matrix
     strategy:
       fail-fast: false

.github/workflows/test-wheel-windows.yml

@@ -88,6 +88,7 @@ jobs:
 
   test:
     name: Python ${{ matrix.PY_VER }}, CUDA ${{ matrix.CUDA_VER }} (${{ (matrix.LOCAL_CTK == '1' && 'local') || 'wheels' }}), GPU ${{ matrix.GPU }}${{ matrix.GPU_COUNT != '1' && format(' (x{0})', matrix.GPU_COUNT) || '' }} (${{ matrix.DRIVER_MODE }})${{ matrix.TORCH_VER && format(', {0}', matrix.TORCH_VER) || '' }}${{ matrix.MODE == 'nightly-numba-cuda' && ', latest' || '' }}
+    timeout-minutes: 60
     # The build stage could fail but we want the CI to keep moving.
     needs: compute-matrix
     strategy:

cuda_bindings/examples/2_Concepts_and_Techniques/stream_ordered_allocation.py

@@ -137,7 +137,7 @@ def stream_ordered_allocation_post_sync(dev, nelem, a, b, c):
             threshold_val,
         )
     )
-    # Record teh start event
+    # Record the start event
     check_cuda_errors(cudart.cudaEventRecord(start, stream))
     for _i in range(MAX_ITER):
         d_a = check_cuda_errors(cudart.cudaMallocAsync(num_bytes, stream))

cuda_core/cuda/core/system/_system.pyx

@@ -11,6 +11,8 @@
 CUDA_BINDINGS_NVML_IS_COMPATIBLE: bool
 
 
+# Please keep in sync with the equivalent implementation in
+# cuda_python_test_helpers/cuda_python_test_helpers/__init__.py.
 cdef bint _detect_wsl():
     try:
         with open("/proc/sys/kernel/osrelease") as f:

cuda_core/tests/memory_ipc/test_leaks.py

@@ -34,7 +34,7 @@ def test_alloc_handle(ipc_memory_resource):
 
 
 def exec_success(obj, number=1):
-    """Succesfully run a child process."""
+    """Successfully run a child process."""
     for _ in range(number):
         process = mp.Process(target=child_main, args=(obj,))
         process.start()
@@ -50,8 +50,8 @@ def child_main(obj, *args):
 
 def exec_launch_failure(obj, number=1):
     """
-    Unsuccesfully try to launch a child process. This fails when
-    after the child starts.
+    Unsuccessfully try to launch a child process. This fails after
+    the child starts.
     """
     for _ in range(number):
         process = mp.Process(target=child_main_bad, args=(obj,))
@@ -68,7 +68,7 @@ def child_main_bad():
 
 def exec_reduce_failure(obj, number=1):
     """
-    Unsuccesfully try to launch a child process. This fails before
+    Unsuccessfully try to launch a child process. This fails before
     the child starts but after the resource-owning object is serialized.
     """
     for _ in range(number):

cuda_python_test_helpers/cuda_python_test_helpers/__init__.py

@@ -15,13 +15,13 @@
 ]
 
 
+# Please keep in sync with the equivalent implementation in
+# cuda_core/cuda/core/system/_system.pyx.
 def _detect_wsl() -> bool:
     data = ""
     with suppress(Exception), open("/proc/sys/kernel/osrelease") as f:
         data = f.read().lower()
-    if "microsoft" in data or "wsl" in data:
-        return True
-    return any(os.environ.get(k) for k in ("WSL_DISTRO_NAME", "WSL_INTEROP"))
+    return "microsoft" in data or "wsl" in data
 
 
 IS_WSL: bool = _detect_wsl()