Allow Pynvjitlinker Enabled via Config at Runtime

numba_cuda/numba/cuda/cudadrv/driver.py

@@ -42,6 +42,11 @@
 from .linkable_code import LinkableCode, LTOIR, Fatbin, Object
 from numba.cuda.cudadrv import enums, drvapi, nvrtc
 
+try:
+    from pynvjitlink.api import NvJitLinker, NvJitLinkError
+except ImportError:
+    NvJitLinker, NvJitLinkError = None, None
+
 USE_NV_BINDING = config.CUDA_USE_NVIDIA_BINDING
 
 if USE_NV_BINDING:
@@ -92,20 +97,6 @@ def _readenv(name, ctor, default):
 if not hasattr(config, "CUDA_ENABLE_PYNVJITLINK"):
     config.CUDA_ENABLE_PYNVJITLINK = ENABLE_PYNVJITLINK
 
-if ENABLE_PYNVJITLINK:
-    try:
-        from pynvjitlink.api import NvJitLinker, NvJitLinkError
-    except ImportError:
-        raise ImportError(
-            "Using pynvjitlink requires the pynvjitlink package to be available"
-        )
-
-    if config.CUDA_ENABLE_MINOR_VERSION_COMPATIBILITY:
-        raise ValueError(
-            "Can't set CUDA_ENABLE_MINOR_VERSION_COMPATIBILITY and "
-            "CUDA_ENABLE_PYNVJITLINK at the same time"
-        )
-
 
 def make_logger():
     logger = logging.getLogger(__name__)
@@ -3061,6 +3052,17 @@ def __init__(
         lto=False,
         additional_flags=None,
     ):
+        if NvJitLinker is None:
+            raise ImportError(
+                "Using pynvjitlink requires the pynvjitlink package to be "
+                "available"
+            )
+
+        if config.CUDA_ENABLE_MINOR_VERSION_COMPATIBILITY:
+            raise ValueError(
+                "Can't set CUDA_ENABLE_MINOR_VERSION_COMPATIBILITY and "
+                "CUDA_ENABLE_PYNVJITLINK at the same time"
+            )
 
         if cc is None:
             raise RuntimeError("PyNvJitLinker requires CC to be specified")

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

@@ -2,16 +2,18 @@
 from numba.cuda.testing import skip_on_cudasim
 from numba.cuda.testing import CUDATestCase
 from numba.cuda.cudadrv.driver import PyNvJitLinker
+from numba.cuda import get_current_device
+
+from numba import cuda
+from numba import config
+from numba.tests.support import run_in_subprocess, override_config
 
 import itertools
 import os
 import io
 import contextlib
 import warnings
 
-from numba.cuda import get_current_device
-from numba import cuda
-from numba import config
 
 TEST_BIN_DIR = os.getenv("NUMBA_CUDA_TEST_BIN_DIR")
 if TEST_BIN_DIR:
@@ -251,5 +253,61 @@ def kernel():
                 pass
 
 
+class TestLinkerUsage(CUDATestCase):
+    """Test that whether pynvjitlink can be enabled by both environment variable
+    and modification of config at runtime.
+    """
+    def test_linker_enabled_envvar(self):
+        # Linkable code is only supported via pynvjitlink
+        src = """if 1:
+        import os
+        from numba import cuda
+
+        TEST_BIN_DIR = os.getenv("NUMBA_CUDA_TEST_BIN_DIR")
+        if TEST_BIN_DIR:
+            test_device_functions_cubin = os.path.join(
+                TEST_BIN_DIR, "test_device_functions.cubin"
+            )
+        print(TEST_BIN_DIR)
+        files = (
+            test_device_functions_cubin,
+        )
+        for lto in [True, False]:
+            for file in files:
+                sig = "uint32(uint32, uint32)"
+                add_from_numba = cuda.declare_device("add_from_numba", sig)
+
+                @cuda.jit(link=[file], lto=lto)
+                def kernel(result):
+                    result[0] = add_from_numba(1, 2)
+
+                result = cuda.device_array(1)
+                kernel[1, 1](result)
+                assert result[0] == 3
+        """
+        env = os.environ.copy()
+        env['NUMBA_CUDA_ENABLE_PYNVJITLINK'] = "1"
+        print(env['NUMBA_CUDA_TEST_BIN_DIR'])
+        run_in_subprocess(src, env=env)
+
+    def test_linker_enabled_config(self):
+        with override_config("CUDA_ENABLE_PYNVJITLINK", True):
+            files = (
+                test_device_functions_cubin,
+            )
+            for lto in [True, False]:
+                for file in files:
+                    sig = "uint32(uint32, uint32)"
+                    add_from_numba = cuda.declare_device("add_from_numba", sig)
+
+                    @cuda.jit(link=[file], lto=lto)
+                    def kernel(result):
+                        result[0] = add_from_numba(1, 2)
+
+                    result = cuda.device_array(1)
+                    kernel[1, 1](result)
+                    assert result[0] == 3
+
+
 if __name__ == "__main__":
     unittest.main()