ref(upstream): FULL TREE — Deep-Spark xllm (1470) + ds_vllm csrc/models (703)

Replaces cherry-picked upstream_ref with complete source trees.

xllm/ — Iluvatar official C++ inference engine (15MB, 1470 files)
  Complete: kernels → layers → models → runtime → scheduler → api
  Excluded: .git, binary images, third_party submodule checkouts

ds_vllm/ — Iluvatar official vllm fork (8MB, 703 files)
  Included: csrc/ (ALL CUDA kernels), fused_moe/, qwen3_5 model, _custom_ops
  Excluded: tests, benchmarks, docs, examples (not needed for reference)

Critical call chains now fully traceable:
  MoE: moe_topk_softmax_kernels.cuh → ixformer.h → fused_moe.cpp → layer
  GDN: qwen3_gated_delta_net_base.cpp → qwen3_5_gated_delta_net.cpp
  Attention: ixformer.h → xllm_paged_attention → attention.cpp

upstream_ref/xllm/xllm/server/xllm_server.h

@@ -0,0 +1,76 @@
+/* Copyright 2025 The xLLM Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    https://github.com/jd-opensource/xllm/blob/main/LICENSE
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+#pragma once
+
+#include "api_service/api_service.h"
+#include "core/distributed_runtime/collective_service.h"
+#include "core/distributed_runtime/disagg_pd_service.h"
+#include "core/distributed_runtime/pd_ooc_service.h"
+#include "core/distributed_runtime/worker_service.h"
+#include "core/framework/xtensor/xtensor_dist_service.h"
+
+namespace xllm {
+
+class XllmServer final {
+ public:
+  XllmServer();
+  ~XllmServer();
+
+  bool start(std::unique_ptr<APIService> api_service);
+  bool start(std::unique_ptr<DisaggPDService> disagg_pd_service);
+  bool start(std::unique_ptr<PDOOCService> pd_ooc_service);
+  bool start(std::shared_ptr<CollectiveService> service,
+             const std::string& addr,
+             const std::string& server_name);
+  bool start(std::shared_ptr<WorkerService> service, const std::string& addr);
+  bool start(std::shared_ptr<XTensorDistService> service,
+             const std::string& addr);
+
+  void run();
+  void stop();
+
+  bool has_initialized() const { return has_initialized_; }
+  std::string listen_address() const { return listen_address_; }
+  int listen_port() const { return listen_port_; }
+
+ private:
+  DISALLOW_COPY_AND_ASSIGN(XllmServer);
+  bool create_server(google::protobuf::Service* service,
+                     const std::string& addr,
+                     int port,
+                     const std::string& server_name);
+
+ private:
+  bool has_initialized_ = false;
+  int listen_port_ = -1;
+  std::string listen_address_;
+  std::unique_ptr<brpc::Server> server_;
+  std::unique_ptr<std::thread> running_thread_;
+};
+
+class XllmServerFactory {
+ public:
+  static std::unique_ptr<XllmServer> create_xllm_server() {
+    return std::make_unique<XllmServer>();
+  }
+
+ private:
+  XllmServerFactory() = default;
+  ~XllmServerFactory() = default;
+  DISALLOW_COPY_AND_ASSIGN(XllmServerFactory);
+};
+
+}  // namespace xllm