[core] Support custom ascendc kernels in vllm-ascend (#233)

This PR add custom ascendc kernel rotary_embedding support in
vllm-ascend, related CMakeLists and setuptools is also added in this PR.

Related: https://github.com/vllm-project/vllm-ascend/issues/156

---------

Signed-off-by: ganyi <pleaplusone.gy@gmail.com>

vllm_ascend/envs.py

@@ -0,0 +1,25 @@
+import os
+from typing import Any, Callable, Dict
+
+env_variables: Dict[str, Callable[[], Any]] = {
+    # max compile thread num
+    "MAX_JOBS": lambda: os.getenv("MAX_JOBS", None),
+    "CMAKE_BUILD_TYPE": lambda: os.getenv("CMAKE_BUILD_TYPE"),
+    "COMPILE_CUSTOM_KERNELS":
+    lambda: os.getenv("COMPILE_CUSTOM_KERNELS", None),
+    # If set, vllm-ascend will print verbose logs during compliation
+    "VERBOSE": lambda: bool(int(os.getenv('VERBOSE', '0'))),
+    "ASCEND_HOME_PATH": lambda: os.getenv("ASCEND_HOME_PATH", None),
+    "LD_LIBRARY_PATH": lambda: os.getenv("LD_LIBRARY_PATH", None),
+}
+
+
+def __getattr__(name: str):
+    # lazy evaluation of environment variables
+    if name in env_variables:
+        return env_variables[name]()
+    raise AttributeError(f"module {__name__!r} has no attribute {name!r}")
+
+
+def __dir__():
+    return list(env_variables.keys())

vllm_ascend/platform.py

@@ -15,6 +15,7 @@
 # limitations under the License.
 #
 
+import logging
 import os
 from typing import TYPE_CHECKING, Optional, Tuple
 
@@ -23,6 +24,19 @@ import torch_npu  # noqa: F401
 import vllm.envs as envs
 from vllm.config import CompilationLevel
 from vllm.logger import init_logger
+
+try:
+    # register custom ops into torch_library here
+    import vllm_ascend.vllm_ascend_C  # type: ignore  # noqa: F401
+
+except ImportError as e:
+    if not str(
+            e
+    ) == "dynamic module does not define module export function (PyInit_vllm_ascend_C)":
+        logging.warning(
+            "Warning: Failed to register custom ops, all custom ops will be disabled"
+        )
+
 from vllm.platforms import Platform, PlatformEnum
 
 if TYPE_CHECKING: