[Chore] Rename SchedulerConfig.chunked_prefill_enabled (#28735)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>

tests/v1/core/test_scheduler.py

@@ -2282,7 +2282,6 @@ def _validate_chunked_prefill_settings_for_encoder_decoder(
 ) -> None:
     """Validate chunked prefill settings in the scheduler config for
     encoder-decoder models."""
-    assert scheduler_config.chunked_prefill_enabled is expect_enabled
     assert scheduler_config.enable_chunked_prefill is expect_enabled
     if is_encoder_decoder:
         # Encoder-decoder models should automatically disable chunked multimodal

tests/v1/e2e/test_spec_decode.py

@@ -272,7 +272,7 @@ def test_speculators_model_integration(
 
 
 @pytest.mark.parametrize(
-    ["model_setup", "mm_enabled", "chunked_prefill_enabled"],
+    ["model_setup", "mm_enabled", "enable_chunked_prefill"],
     [
         (("eagle3", "Qwen/Qwen3-8B", "AngelSlim/Qwen3-8B_eagle3", 1), False, False),
         pytest.param(
@@ -358,7 +358,7 @@ def test_eagle_correctness(
     sampling_config: SamplingParams,
     model_setup: tuple[str, str, str, int],
     mm_enabled: bool,
-    chunked_prefill_enabled: bool,
+    enable_chunked_prefill: bool,
     attn_backend: str,
 ):
     if attn_backend == "TREE_ATTN":
@@ -396,9 +396,7 @@ def test_eagle_correctness(
 
         method, model_name, spec_model_name, tp_size = model_setup
         max_model_len = 2048
-        max_num_batched_tokens = max_model_len
-        if chunked_prefill_enabled:
-            max_num_batched_tokens = 128
+        max_num_batched_tokens = 128 if enable_chunked_prefill else max_model_len
 
         ref_llm = LLM(
             model=model_name, max_model_len=max_model_len, tensor_parallel_size=tp_size
@@ -420,7 +418,7 @@ def test_eagle_correctness(
             },
             max_model_len=max_model_len,
             max_num_batched_tokens=max_num_batched_tokens,
-            enable_chunked_prefill=chunked_prefill_enabled,
+            enable_chunked_prefill=enable_chunked_prefill,
         )
         spec_outputs = spec_llm.chat(test_prompts, sampling_config)
         matches = 0

tests/v1/engine/test_engine_core.py

@@ -571,7 +571,7 @@ def test_encoder_instance_zero_kv_cache(
         )
 
         # Check 5: Verify chunked prefill is disabled
-        assert not vllm_config.scheduler_config.chunked_prefill_enabled, (
+        assert not vllm_config.scheduler_config.enable_chunked_prefill, (
             "Encoder instance should disable chunked prefill (no KV cache)"
         )