fix tune failed for DynamicRnn

mindspore/_extends/parallel_compile/tbe_compiler/tbe_adapter.py

@@ -32,7 +32,7 @@ from te_fusion.parallel_compilation import init_multi_process_env, start_ga_mult
     get_finished_compilation_task
 
 from .tbe_helper import get_soc_info, assemble_op_args, get_compute_op_list, get_options_info, get_fuzz_build_info, \
-    BuildType, adjust_custom_op_info
+    BuildType, adjust_custom_op_info, pack_op_args
 from .tbe_job import TbeJob, JobStatus
 
 PLATFORM_FLAG = ["Ascend310", "Ascend910", "Hi3796CV300ES", "Ascend710", "Ascend610", "Hi3796CV300CS", "SD3403"]
@@ -516,8 +516,9 @@ def rl_tune_single_op(job: TbeJob):
     tune_op_module_name = op_module_name + "@" + py_module_path
     base_kernel = job.content["SocInfo"]["op_debug_dir"] + "/kernel_meta/" + op_kernel_name + ".o"
     from schedule_search.rl_online_tune import dispatch_single_tune_task
+    pack_args = pack_op_args(inputs, outputs, attrs)
     res = dispatch_single_tune_task(job.source_id, job.id, l1_size, base_kernel, op_kernel_name, full_name,
-                                    tune_op_module_name, op_func_name, op_type, (inputs, outputs, attrs))
+                                    tune_op_module_name, op_func_name, op_type, pack_args)
     res = _process_rl_tune_result(job, res)
     return res
 

mindspore/_extends/parallel_compile/tbe_compiler/tbe_helper.py

@@ -232,3 +232,11 @@ def adjust_custom_op_info(compute_op_info):
         module_name, _ = os.path.splitext(file_name)
         compute_op_info["py_module_path"] = py_module_path
         compute_op_info["module_name"] = module_name
+
+
+def pack_op_args(inputs, outputs, attrs):
+    """
+    flatten inputs outputs attrs
+    """
+    op_args = (inputs, outputs, attrs)
+    return [item for arg in op_args for item in arg]

mindspore/_extends/parallel_compile/tbe_compiler/tuner.py

@@ -312,9 +312,10 @@ class TbeTuner:
         else:
             job_type = RL_ONLINE
         graph_id = 0
-        l1size = 0  # todo need to verify
+        l1size = 0
+        pack_op_args = [item for arg in op_args for item in arg]
         ret = dispatch_single_tune_task(graph_id, task_id, l1size, base_kernel, kernel_name, full_name,
-                                        op_module_name + "@" + op_module_name, op_type, op_type, op_args)
+                                        op_module_name + "@" + op_module_name, op_type, op_type, pack_op_args)
 
         self.module_list[op_module_name] = 1
         self.fusion_need_sync += 1