Upgrade to new vllm commit (#3719)

### What this PR does / why we need it?
Upgrade to new vllm commit:
c9461e05a4

- Fix many imports, caused by
https://github.com/vllm-project/vllm/pull/26908
- Fix import ```sha256```, caused by
https://github.com/vllm-project/vllm/pull/27169
- Remove ```SchedulerConfig.send_delta_data```, caused by
https://github.com/vllm-project/vllm/pull/27142
- Fix ```FusedMoE``` because of dual stream execution, caused by
https://github.com/vllm-project/vllm/pull/26440

### Does this PR introduce _any_ user-facing change?
N/A

### How was this patch tested?
CI passed with new added/existing test.


- vLLM version: v0.11.0rc3
- vLLM main:
17c540a993

---------

Signed-off-by: MengqingCao <cmq0113@163.com>
Signed-off-by: Icey <1790571317@qq.com>
Co-authored-by: MengqingCao <cmq0113@163.com>

tests/ut/core/test_schedule_config.py

@@ -78,21 +78,6 @@ class TestAscendSchedulerConfig(TestBase):
             str(context.exception),
         )
 
-    def test_not_implemented_send_delta_data(self):
-        with self.assertRaises(NotImplementedError) as context:
-            AscendSchedulerConfig.initialize_from_config(
-                self.basic_scheduler_config,
-                AscendSchedulerConfig(
-                    send_delta_data=True,
-                    max_num_batched_tokens=2048,
-                    max_model_len=2048,
-                ),
-            )
-        self.assertIn(
-            "currently AscendScheduler doesn't support send_delta_data",
-            str(context.exception),
-        )
-
     def test_no_override(self):
         ascend_config = AscendSchedulerConfig.initialize_from_config(
             self.basic_scheduler_config, {})

tests/ut/core/test_scheduler.py

@@ -9,7 +9,6 @@ from vllm.config import (CacheConfig, KVTransferConfig, ModelConfig,
 from vllm.multimodal.inputs import (MultiModalFeatureSpec,
                                     MultiModalKwargsItem, PlaceholderRange)
 from vllm.sampling_params import SamplingParams
-from vllm.utils import sha256
 from vllm.v1.core.kv_cache_utils import (get_request_block_hasher,
                                          init_none_hash)
 from vllm.v1.core.sched.output import SchedulerOutput
@@ -24,6 +23,11 @@ from vllm_ascend.core.scheduler import AscendScheduler
 from vllm_ascend.core.scheduler_dynamic_batch import SchedulerDynamicBatch
 from vllm_ascend.utils import vllm_version_is
 
+if vllm_version_is("0.11.0"):
+    from vllm.utils import sha256
+else:
+    from vllm.utils.hashing import sha256
+
 EOS_TOKEN_ID = 50256
 MODEL = "Qwen3-0.6B"
 ENABLE_PREFIX_CACHING = None

tests/ut/kv_connector/test_mooncake_connector.py

@@ -12,7 +12,13 @@ from unittest.mock import MagicMock, patch
 
 import msgspec
 import zmq
-from vllm.utils import make_zmq_path
+
+from vllm_ascend.utils import vllm_version_is
+
+if vllm_version_is("0.11.0"):
+    from vllm.utils import make_zmq_path
+else:
+    from vllm.utils.network_utils import make_zmq_path
 
 fake_engine = types.ModuleType("mooncake.engine")
 fake_engine.TransferEngine = MagicMock()  # type: ignore[attr-defined]

tests/ut/kv_connector/utils.py

@@ -10,7 +10,6 @@ import torch
 from vllm import SamplingParams
 from vllm.config import (CacheConfig, DeviceConfig, KVTransferConfig,
                          ModelConfig, SchedulerConfig, VllmConfig)
-from vllm.utils import sha256
 from vllm.v1.core.kv_cache_utils import (get_request_block_hasher,
                                          init_none_hash)
 from vllm.v1.core.sched.scheduler import Scheduler
@@ -22,6 +21,11 @@ from vllm.v1.structured_output import StructuredOutputManager
 
 from vllm_ascend.utils import vllm_version_is
 
+if vllm_version_is("0.11.0"):
+    from vllm.utils import sha256
+else:
+    from vllm.utils.hashing import sha256
+
 EOS_TOKEN_ID = 50256
 os.environ["VLLM_USE_V1"] = "1"
 

tests/ut/model_loader/netloader/test_netloader.py

@@ -22,6 +22,7 @@ import torch
 from torch import nn
 
 from vllm_ascend.model_loader.netloader.netloader import ModelNetLoaderElastic
+from vllm_ascend.utils import vllm_version_is
 
 
 class DummyDeviceConfig:
@@ -173,7 +174,11 @@ def test_load_model_elastic_success(mock_logger, monkeypatch, tmp_path):
         "vllm_ascend.model_loader.netloader.netloader.process_weights_after_loading",
         lambda *a, **k: None)
     # patch get_ip
-    monkeypatch.setattr("vllm.utils.get_ip", lambda: "127.0.0.1")
+    if vllm_version_is("0.11.0"):
+        monkeypatch.setattr("vllm.utils.get_ip", lambda: "127.0.0.1")
+    else:
+        monkeypatch.setattr("vllm.utils.network_utils.get_ip",
+                            lambda: "127.0.0.1")
     # patch find_free_port
     monkeypatch.setattr(
         "vllm_ascend.model_loader.netloader.netloader.find_free_port",

tests/ut/worker/test_input_batch.py

@@ -20,14 +20,19 @@ import numpy as np
 import pytest
 import torch
 from vllm.sampling_params import SamplingParams
-from vllm.utils import make_tensor_with_pad
 from vllm.v1.pool.metadata import PoolingMetadata
 from vllm.v1.sample.logits_processor import LogitsProcessors
 from vllm.v1.sample.metadata import SamplingMetadata
 
+from vllm_ascend.utils import vllm_version_is
 from vllm_ascend.worker.block_table import BlockTable, MultiGroupBlockTable
 from vllm_ascend.worker.npu_input_batch import CachedRequestState, InputBatch
 
+if vllm_version_is("0.11.0"):
+    from vllm.utils import make_tensor_with_pad
+else:
+    from vllm.utils.torch_utils import make_tensor_with_pad
+
 VOCAB_SIZE = 1024
 NUM_OUTPUT_TOKENS = 20
 MAX_PROMPT_SIZE = 100

tests/ut/worker/test_worker_v1.py

@@ -5,6 +5,7 @@ import torch
 from vllm.config import CacheConfig, ModelConfig, ParallelConfig, VllmConfig
 
 from tests.ut.base import TestBase
+from vllm_ascend.utils import vllm_version_is
 
 
 class TestNPUWorker(TestBase):
@@ -178,15 +179,26 @@ class TestNPUWorker(TestBase):
         # Create NPUWorker instance
         from vllm_ascend.worker.worker_v1 import NPUWorker
 
-        with patch("vllm.utils.STR_DTYPE_TO_TORCH_DTYPE",
-                   {"float32": torch.float32}):
-            worker = NPUWorker(
-                vllm_config=self.vllm_config_mock,
-                local_rank=self.local_rank,
-                rank=self.rank,
-                distributed_init_method=self.distributed_init_method,
-                is_driver_worker=self.is_driver_worker,
-            )
+        if vllm_version_is("0.11.0"):
+            with patch("vllm.utils.STR_DTYPE_TO_TORCH_DTYPE",
+                       {"float32": torch.float32}):
+                worker = NPUWorker(
+                    vllm_config=self.vllm_config_mock,
+                    local_rank=self.local_rank,
+                    rank=self.rank,
+                    distributed_init_method=self.distributed_init_method,
+                    is_driver_worker=self.is_driver_worker,
+                )
+        else:
+            with patch("vllm.utils.torch_utils.STR_DTYPE_TO_TORCH_DTYPE",
+                       {"float32": torch.float32}):
+                worker = NPUWorker(
+                    vllm_config=self.vllm_config_mock,
+                    local_rank=self.local_rank,
+                    rank=self.rank,
+                    distributed_init_method=self.distributed_init_method,
+                    is_driver_worker=self.is_driver_worker,
+                )
 
         # Verify cache_dtype is set to custom value
         self.assertEqual(worker.cache_dtype, torch.float32)