Reapply [Attention] Refactor check_and_update_config (#35122)

Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>

vllm/v1/executor/multiproc_executor.py

@@ -42,6 +42,7 @@ from vllm.distributed.parallel_state import (
 )
 from vllm.envs import enable_envs_cache
 from vllm.logger import init_logger
+from vllm.platforms import current_platform
 from vllm.tracing import instrument, maybe_init_worker_tracer
 from vllm.utils.network_utils import (
     get_distributed_init_method,
@@ -617,6 +618,9 @@ class WorkerProc:
             )
             self.worker.load_model()
 
+        # Set block size based on the attention backends
+        current_platform.update_block_size_for_backend(vllm_config)
+
         # Initialize message queues after init_device() since multi-node setups
         # (nnodes_within_dp > 1) require distributed groups to be initialized
         self._init_message_queues(input_shm_handle, vllm_config)