[Bugfix] Fix broken CI (#1848)

### What this PR does / why we need it?
- Fix broken commit by
[#20927](vllm-project/vllm#20927)
- Fix broken commit by
[#20466](vllm-project/vllm#20466)
- TODO: more fully adapt to the upstream reconstruction, let's first
make CI happy

- vLLM version: v0.9.2
- vLLM main:
vllm-project/vllm@11dfdf2

---------

Signed-off-by: wangli <wangli858794774@gmail.com>

Author: Potabk

vllm_ascend/pool/__init__.py

@@ -59,6 +59,7 @@
                                         KVCacheSpec)
 from vllm.v1.outputs import (EMPTY_MODEL_RUNNER_OUTPUT, LogprobsTensors,
                              ModelRunnerOutput)
+from vllm.v1.pool.metadata import PoolingMetadata
 from vllm.v1.sample.metadata import SamplingMetadata
 from vllm.v1.sample.sampler import Sampler
 from vllm.v1.spec_decode.metadata import SpecDecodeMetadata
@@ -76,7 +77,6 @@
 from vllm_ascend.attention.mla_v1 import (AscendMLAMetadata,
                                           CommonAttentionMetadata)
 from vllm_ascend.platform import NPUPlatform
-from vllm_ascend.pool.metadata import PoolingMetadata
 from vllm_ascend.sample.rejection_sampler import AscendRejectionSampler
 from vllm_ascend.utils import (ACL_FORMAT_FRACTAL_ND, ACL_FORMAT_FRACTAL_NZ,
                                ProfileExecuteDuration,
@@ -571,7 +571,10 @@ def get_eagle_atten_dict(
 
         # OPTIMIZATION: Start copying the block table first.
         # This way, we can overlap the copy with the following CPU operations.
-        self.input_batch.block_table.commit(num_reqs)
+        if vllm_version_is("0.9.2"):
+            self.input_batch.block_table.commit(num_reqs)
+        else:
+            self.input_batch.block_table.commit_block_table(num_reqs)
 
         # Get the number of scheduled tokens for each request.
         req_ids = self.input_batch.req_ids
@@ -902,7 +905,10 @@ def _process_reqs(
 
         # OPTIMIZATION: Start copying the block table first.
         # This way, we can overlap the copy with the following CPU operations.
-        self.input_batch.block_table.commit(num_reqs)
+        if vllm_version_is("0.9.2"):
+            self.input_batch.block_table.commit(num_reqs)
+        else:
+            self.input_batch.block_table.commit_block_table(num_reqs)
 
         # Get the number of scheduled tokens for each request.
         # TODO: The Python loop can be slow. Optimize.

vllm_ascend/pool/metadata.py

@@ -28,14 +28,13 @@
 from vllm.sampling_params import SamplingParams, SamplingType
 from vllm.utils import swap_dict_values
 from vllm.v1.outputs import LogprobsTensors
+from vllm.v1.pool.metadata import PoolingMetadata
 from vllm.v1.sample.logits_processor import init_builtin_logitsprocs
 from vllm.v1.sample.metadata import SamplingMetadata
 from vllm.v1.spec_decode.utils import is_spec_decode_unsupported
 from vllm.v1.utils import copy_slice
 from vllm.v1.worker.block_table import MultiGroupBlockTable
 
-from vllm_ascend.pool.metadata import PoolingMetadata
-
 _SAMPLING_EPS = 1e-5