Merge branch 'main' into bandit_pre-commit_workflow_sync

Author: rwgk

.github/workflows/build-wheel.yml

@@ -96,7 +96,7 @@ jobs:
       - name: Check cuda.pathfinder wheel
         if: ${{ strategy.job-index == 0 && inputs.host-platform == 'linux-64' }}
         run: |
-          twine check cuda_pathfinder/*.whl
+          twine check --strict cuda_pathfinder/*.whl
 
       - name: Upload cuda.pathfinder build artifacts
         if: ${{ strategy.job-index == 0 && inputs.host-platform == 'linux-64' }}
@@ -132,7 +132,7 @@ jobs:
 
       - name: Check cuda.core wheel
         run: |
-          twine check ${{ env.CUDA_CORE_ARTIFACTS_DIR }}/*.whl
+          twine check --strict ${{ env.CUDA_CORE_ARTIFACTS_DIR }}/*.whl
 
       - name: Upload cuda.core build artifacts
         uses: actions/upload-artifact@ea165f8d65b6e75b540449e92b4886f43607fa02  # v4.6.2
@@ -179,7 +179,7 @@ jobs:
 
       - name: Check cuda.bindings wheel
         run: |
-          twine check ${{ env.CUDA_BINDINGS_ARTIFACTS_DIR }}/*.whl
+          twine check --strict ${{ env.CUDA_BINDINGS_ARTIFACTS_DIR }}/*.whl
 
       - name: Upload cuda.bindings build artifacts
         uses: actions/upload-artifact@ea165f8d65b6e75b540449e92b4886f43607fa02  # v4.6.2
@@ -194,7 +194,7 @@ jobs:
         run: |
           pushd cuda_python
           pip wheel -v --no-deps .
-          twine check *.whl
+          twine check --strict *.whl
           popd
 
       - name: List the cuda-python artifacts directory

.github/workflows/codeql.yml

@@ -31,13 +31,13 @@ jobs:
       uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8  # v5.0.0
 
     - name: Initialize CodeQL
-      uses: github/codeql-action/init@3c3833e0f8c1c83d449a7478aa59c036a9165498  # v3.29.11
+      uses: github/codeql-action/init@2d92b76c45b91eb80fc44c74ce3fce0ee94e8f9d  # v3.30.0
       with:
         languages: ${{ matrix.language }}
         build-mode: ${{ matrix.build-mode }}
         queries: security-extended
 
     - name: Perform CodeQL Analysis
-      uses: github/codeql-action/analyze@3c3833e0f8c1c83d449a7478aa59c036a9165498  # v3.29.11
+      uses: github/codeql-action/analyze@2d92b76c45b91eb80fc44c74ce3fce0ee94e8f9d  # v3.30.0
       with:
         category: "/language:${{matrix.language}}"

.gitignore

@@ -29,10 +29,6 @@ cuda_bindings/cuda/bindings/_bindings/cynvrtc.pxd
 cuda_bindings/cuda/bindings/_bindings/cynvrtc.pyx
 cuda_bindings/cuda/bindings/_internal/nvjitlink.pyx
 cuda_bindings/cuda/bindings/_internal/nvvm.pyx
-cuda_bindings/cuda/bindings/_lib/cyruntime/cyruntime.pxd
-cuda_bindings/cuda/bindings/_lib/cyruntime/cyruntime.pyx
-cuda_bindings/cuda/bindings/_lib/cyruntime/utils.pxd
-cuda_bindings/cuda/bindings/_lib/cyruntime/utils.pyx
 cuda_bindings/cuda/bindings/_lib/utils.pxd
 cuda_bindings/cuda/bindings/_lib/utils.pyx
 cuda_bindings/cuda/bindings/cydriver.pxd

cuda_bindings/cuda/bindings/_bindings/cydriver.pyx.in

@@ -12,9 +12,10 @@ cimport cuda.bindings._lib.dlfcn as dlfcn
 from libc.stdint cimport intptr_t
 import os
 import sys
-import threading
 cimport cuda.bindings._bindings.loader as loader
-cdef object __symbol_lock = threading.RLock()
+import threading
+
+cdef object __symbol_lock = threading.Lock()
 cdef bint __cuPythonInit = False
 {{if 'cuGetErrorString' in found_functions}}cdef void *__cuGetErrorString = NULL{{endif}}
 {{if 'cuGetErrorName' in found_functions}}cdef void *__cuGetErrorName = NULL{{endif}}
@@ -510,7 +511,7 @@ cdef int _cuPythonInit() except -1 nogil:
             {{else}}
             path = 'libcuda.so.1'
             {{endif}}
-    
+
         {{if 'Windows' == platform.system()}}
         LOAD_LIBRARY_SEARCH_SYSTEM32 = 0x00000800
         try:
@@ -522,7 +523,7 @@ cdef int _cuPythonInit() except -1 nogil:
         if (handle == NULL):
             raise RuntimeError('Failed to dlopen ' + path)
         {{endif}}
-    
+
         # Get latest __cuGetProcAddress_v2
         global __cuGetProcAddress_v2
         {{if 'Windows' == platform.system()}}
@@ -533,7 +534,7 @@ cdef int _cuPythonInit() except -1 nogil:
         {{else}}
         __cuGetProcAddress_v2 = dlfcn.dlsym(handle, 'cuGetProcAddress_v2')
         {{endif}}
-    
+
         # Load using cuGetProcAddress if available
         if __cuGetProcAddress_v2 != NULL:
             _F_cuGetProcAddress_v2 = <__cuGetProcAddress_v2_T>__cuGetProcAddress_v2
@@ -2760,10 +2761,9 @@ cdef int _cuPythonInit() except -1 nogil:
             global __cuGraphicsVDPAURegisterOutputSurface
             _F_cuGetProcAddress_v2('cuGraphicsVDPAURegisterOutputSurface', &__cuGraphicsVDPAURegisterOutputSurface, 3010, CU_GET_PROC_ADDRESS_DEFAULT, NULL)
             {{endif}}
-
             __cuPythonInit = True
             return 0
-    
+
         {{if 'Windows' == platform.system()}}
         # Load using win32GetAddr
         if usePTDS:
@@ -8877,7 +8877,6 @@ cdef int _cuPythonInit() except -1 nogil:
         __cuGraphicsVDPAURegisterOutputSurface = dlfcn.dlsym(handle, 'cuGraphicsVDPAURegisterOutputSurface')
         {{endif}}
         {{endif}}
-
         __cuPythonInit = True
         return 0
 
@@ -8886,7 +8885,6 @@ cdef int _cuPythonInit() except -1 nogil:
 cdef inline int cuPythonInit() except -1 nogil:
     if __cuPythonInit:
         return 0
-    
     return _cuPythonInit()
 
 {{if 'cuGetErrorString' in found_functions}}

cuda_bindings/cuda/bindings/_bindings/cynvrtc.pyx.in

@@ -43,6 +43,7 @@ cdef bint __cuPythonInit = False
 cdef int _cuPythonInit() except -1 nogil:
     global __cuPythonInit
 
+    # Load library
     with gil, __symbol_lock:
         {{if 'Windows' == platform.system()}}
         handle = load_nvidia_dynamic_lib("nvrtc")._handle_uint
@@ -317,8 +318,8 @@ cdef int _cuPythonInit() except -1 nogil:
         global __nvrtcSetFlowCallback
         __nvrtcSetFlowCallback = dlfcn.dlsym(handle, 'nvrtcSetFlowCallback')
         {{endif}}
-        {{endif}}
 
+        {{endif}}
         __cuPythonInit = True
         return 0
 
@@ -327,7 +328,6 @@ cdef int _cuPythonInit() except -1 nogil:
 cdef inline int cuPythonInit() except -1 nogil:
     if __cuPythonInit:
         return 0
-
     return _cuPythonInit()
 
 {{if 'nvrtcGetErrorString' in found_functions}}

cuda_bindings/cuda/bindings/_bindings/cyruntime.pxd.in

@@ -4,6 +4,8 @@
 # This code was automatically generated with version 13.0.0. Do not modify it directly.
 include "../cyruntime_types.pxi"
 
+include "../_lib/cyruntime/cyruntime.pxd"
+
 {{if 'cudaDeviceReset' in found_functions}}
 
 cdef cudaError_t _cudaDeviceReset() except ?cudaErrorCallRequiresNewerDriver nogil

cuda_bindings/cuda/bindings/_bindings/cyruntime.pyx.in

@@ -13,6 +13,7 @@ cdef bint __usePTDS = False
 cdef int _cudaPythonInit() except -1 nogil:
         global __cudaPythonInit
         global __usePTDS
+
         with gil:
             __usePTDS = os.getenv('CUDA_PYTHON_CUDA_PER_THREAD_DEFAULT_STREAM', default=False)
         __cudaPythonInit = True
@@ -23,7 +24,6 @@ cdef int _cudaPythonInit() except -1 nogil:
 cdef inline int cudaPythonInit() except -1 nogil:
     if __cudaPythonInit:
         return __usePTDS
-
     return _cudaPythonInit()
 
 {{if 'cudaDeviceReset' in found_functions}}
@@ -2671,3 +2671,6 @@ cdef cudaError_t _cudaProfilerStop() except ?cudaErrorCallRequiresNewerDriver no
         return ptds._cudaProfilerStop()
     return cudaProfilerStop()
 {{endif}}
+
+
+include "../_lib/cyruntime/cyruntime.pxi"

cuda_bindings/cuda/bindings/_internal/cufile_linux.pyx

@@ -13,6 +13,7 @@ from cuda.pathfinder import load_nvidia_dynamic_lib
 
 import cython
 
+
 ###############################################################################
 # Extern
 ###############################################################################

cuda_bindings/cuda/bindings/_internal/nvjitlink_linux.pyx

@@ -5,8 +5,8 @@
 # This code was automatically generated across versions from 12.0.1 to 13.0.0. Do not modify it directly.
 
 from libc.stdint cimport intptr_t, uintptr_t
-import threading
 
+import threading
 from .utils import FunctionNotFoundError, NotSupportedError
 
 from cuda.pathfinder import load_nvidia_dynamic_lib

cuda_bindings/cuda/bindings/_internal/nvjitlink_windows.pyx

@@ -5,8 +5,8 @@
 # This code was automatically generated across versions from 12.0.1 to 13.0.0. Do not modify it directly.
 
 from libc.stdint cimport intptr_t
-import threading
 
+import threading
 from .utils import FunctionNotFoundError, NotSupportedError
 
 from cuda.pathfinder import load_nvidia_dynamic_lib