[PluggableLayer][2/N] Apply PluggableLayer to linear layers (#33152)

Signed-off-by: whx-sjtu <2952154980@qq.com>

vllm/model_executor/layers/linear.py

@@ -17,7 +17,7 @@ from vllm.distributed import (
     tensor_model_parallel_all_reduce,
 )
 from vllm.logger import init_logger
-from vllm.model_executor.custom_op import CustomOp
+from vllm.model_executor.custom_op import PluggableLayer
 from vllm.model_executor.layers.quantization.base_config import (
     QuantizationConfig,
     QuantizeMethodBase,
@@ -239,7 +239,7 @@ class UnquantizedLinearMethod(LinearMethodBase):
         return dispatch_unquantized_gemm()(layer, x, layer.weight, bias)
 
 
-class LinearBase(CustomOp):
+class LinearBase(PluggableLayer):
     """Base linear layer.
 
     Args:
@@ -294,7 +294,7 @@ class LinearBase(CustomOp):
 
 
 # --8<-- [start:replicated_linear]
-@CustomOp.register("replicated_linear")
+@PluggableLayer.register("replicated_linear")
 class ReplicatedLinear(LinearBase):
     """Replicated linear layer.
 
@@ -414,7 +414,7 @@ class ReplicatedLinear(LinearBase):
 
 
 # --8<-- [start:column_parallel_linear]
-@CustomOp.register("column_parallel_linear")
+@PluggableLayer.register("column_parallel_linear")
 class ColumnParallelLinear(LinearBase):
     """Linear layer with column parallelism.
 
@@ -1273,7 +1273,7 @@ class QKVParallelLinear(ColumnParallelLinear):
 
 
 # --8<-- [start:row_parallel_linear]
-@CustomOp.register("row_parallel_linear")
+@PluggableLayer.register("row_parallel_linear")
 class RowParallelLinear(LinearBase):
     """Linear layer with row parallelism.