Allow CUDA VMM feature transport with the Rust frontend (#39347)

python/sglang/srt/arg_groups/serving_hook.py

@@ -851,11 +851,6 @@ def handle_multimodal_feature_transport(server_args: Any):
             raise ValueError(
                 "--mm-feature-transport=cuda_vmm does not support pipeline parallelism."
             )
-        if envs.SGLANG_RUST_SERVER.get():
-            raise ValueError(
-                "--mm-feature-transport=cuda_vmm is not supported with "
-                "SGLANG_RUST_SERVER."
-            )
         pool_budget_mb = envs.SGLANG_MM_FEATURE_CACHE_MB.get()
         handle_kind = "CUDA FABRIC" if cfg.nnodes > 1 else "POSIX FD"
         logger.info(

test/registered/unit/server_args/test_server_args.py

@@ -686,15 +686,19 @@ class TestMultimodalFeatureTransport(CustomTestCase):
             handle_multimodal_feature_transport(server_args)
 
     @override_platform(is_cuda=True)
-    def test_cuda_vmm_rejects_rust_server(self):
+    def test_cuda_vmm_allows_rust_server(self):
         server_args = ServerArgs(model_path="dummy", mm_feature_transport="cuda_vmm")
 
         with (
+            patch.dict(os.environ, {}, clear=False),
             envs.SGLANG_RUST_SERVER.override(True),
-            self.assertRaisesRegex(ValueError, "SGLANG_RUST_SERVER"),
         ):
             handle_multimodal_feature_transport(server_args)
 
+        self.assertEqual(
+            resolution_result(server_args, "mm_feature_transport"), "cuda_vmm"
+        )
+
     @override_platform(is_cuda=True)
     def test_cuda_vmm_rejects_pipeline_parallelism(self):
         server_args = ServerArgs(