Fix Triton HMAC security vulnerabilities (v2)

- Bug 1: Add HMAC integrity check before pickle deserialization in
  Triton handler initialize() method (model.py)
- Bug 2: Replace hardcoded secret key with generate_secret_key() and
  add _hmac_signing() after ONNX exports (triton_builder.py)
- Bug 3: Add secret key validation in _start_triton_server() to reject
  empty/None keys before passing to container (server.py)

Aligns Triton code path with existing HMAC verification patterns used
by TorchServe, MMS, TF Serving, and SMD handlers.

Ticket: P400136088

Author: Pravali Uppugunduri

src/sagemaker/serve/model_server/triton/model.py

@@ -26,10 +26,14 @@ def auto_complete_config(auto_complete_model_config):
     def initialize(self, args: dict) -> None:
         """Placeholder docstring"""
         serve_path = Path(TRITON_MODEL_DIR).joinpath("serve.pkl")
+        metadata_path = Path(TRITON_MODEL_DIR).joinpath("metadata.json")
+
         with open(str(serve_path), mode="rb") as f:
-            inference_spec, schema_builder = cloudpickle.load(f)
+            buffer = f.read()
+        perform_integrity_check(buffer=buffer, metadata_path=str(metadata_path))
 
-        # TODO: HMAC signing for integrity check
+        with open(str(serve_path), mode="rb") as f:
+            inference_spec, schema_builder = cloudpickle.load(f)
 
         self.inference_spec = inference_spec
         self.schema_builder = schema_builder

src/sagemaker/serve/model_server/triton/server.py

@@ -43,11 +43,16 @@ def _start_triton_server(
         env_vars.update(
             {
                 "TRITON_MODEL_DIR": "/models/model",
-                "SAGEMAKER_SERVE_SECRET_KEY": secret_key,
                 "LOCAL_PYTHON": platform.python_version(),
             }
         )
 
+        # Only set SAGEMAKER_SERVE_SECRET_KEY for inference_spec path where
+        # pickle integrity verification is needed. The ONNX path does not
+        # use pickles, so no secret key is required.
+        if secret_key and isinstance(secret_key, str) and secret_key.strip():
+            env_vars["SAGEMAKER_SERVE_SECRET_KEY"] = secret_key
+
         if "cpu" not in image_uri:
             self.container = docker_client.containers.run(
                 image=image_uri,
@@ -146,7 +151,12 @@ def _upload_triton_artifacts(
         env_vars = {
             "SAGEMAKER_TRITON_DEFAULT_MODEL_NAME": "model",
             "TRITON_MODEL_DIR": "/opt/ml/model/model",
-            "SAGEMAKER_SERVE_SECRET_KEY": secret_key,
             "LOCAL_PYTHON": platform.python_version(),
         }
+
+        # Only set SAGEMAKER_SERVE_SECRET_KEY for inference_spec path where
+        # pickle integrity verification is needed.
+        if secret_key and isinstance(secret_key, str) and secret_key.strip():
+            env_vars["SAGEMAKER_SERVE_SECRET_KEY"] = secret_key
+
         return s3_upload_path, env_vars

src/sagemaker/serve/model_server/triton/triton_builder.py

@@ -213,8 +213,8 @@ def _prepare_for_triton(self):
             export_path.mkdir(parents=True)
 
         if self.model:
-            self.secret_key = "dummy secret key for onnx backend"
-
+            # ONNX path: export model to ONNX format for Triton's native ONNX backend.
+            # No pickle is created or loaded at runtime, so no HMAC signing is needed.
             if self._framework == "pytorch":
                 self._export_pytorch_to_onnx(
                     export_path=export_path, model=self.model, schema_builder=self.schema_builder
@@ -457,13 +457,11 @@ def _get_triton_predictor(self, endpoint_name: str, sagemaker_session: Session)
         )
 
     def _save_inference_spec(self) -> None:
-        """Placeholder docstring"""
+        """Save inference specification to pickle file."""
         if self.inference_spec:
             pkl_path = Path(self.model_path).joinpath("model_repository").joinpath("model")
             save_pkl(pkl_path, (self.inference_spec, self.schema_builder))
 
-        return
-
     def _build_for_triton(self):
         """Placeholder docstring"""
         self._validate_for_triton()

tests/unit/sagemaker/serve/model_server/triton/test_triton_builder.py

@@ -67,6 +67,7 @@ def prepare_triton_builder_for_model(self, triton_builder: Triton) -> Triton:
         mock_export = Mock()
         triton_builder._export_pytorch_to_onnx = mock_export
         triton_builder._export_tf_to_onnx = mock_export
+        triton_builder._hmac_signing = Mock()
 
         return triton_builder