fix structure infos conflict in static return_list mode (#43291)

* fix structure infos conflict in static return_list mode. test=develop

python/paddle/fluid/dataloader/dataloader_iter.py

@@ -276,12 +276,12 @@ class _DataLoaderIterSingleProcess(_DataLoaderIterBase):
                         data = self._reader.read_next_list()
                         for i in range(len(data)):
                             data[i] = data[i]._move_to_list()
-                        data = [
-                            _restore_batch(d, s) for d, s in zip(
-                                data, self._structure_infos[:len(self._places)])
+                        structs = [
+                            self._structure_infos.pop(0)
+                            for _ in range(len(self._places))
                         ]
-                        self._structure_infos = self._structure_infos[
-                            len(self._places):]
+                        data = [_restore_batch(d, s) \
+                                for d, s in zip(data, structs)]
                         # static graph organized data on multi-device with list, if
                         # place number is 1, there is only 1 device, extra the data
                         # from list for devices to be compatible with dygraph mode
@@ -750,12 +750,12 @@ class _DataLoaderIterMultiProcess(_DataLoaderIterBase):
                         data = self._reader.read_next_list()
                         for i in range(len(data)):
                             data[i] = data[i]._move_to_list()
-                        data = [
-                            _restore_batch(d, s) for d, s in zip(
-                                data, self._structure_infos[:len(self._places)])
+                        structs = [
+                            self._structure_infos.pop(0)
+                            for _ in range(len(self._places))
                         ]
-                        self._structure_infos = self._structure_infos[
-                            len(self._places):]
+                        data = [_restore_batch(d, s) \
+                                for d, s in zip(data, structs)]
                         # static graph organized data on multi-device with list, if
                         # place number is 1, there is only 1 device, extra the data
                         # from list for devices to be compatible with dygraph mode

python/paddle/fluid/tests/unittests/test_multiprocess_dataloader_static.py

@@ -22,6 +22,7 @@ import unittest
 import multiprocessing
 import numpy as np
 
+import paddle
 import paddle.fluid as fluid
 from paddle.io import Dataset, BatchSampler, DataLoader
 
@@ -182,7 +183,7 @@ class TestStaticDataLoader(unittest.TestCase):
 
 class TestStaticDataLoaderReturnList(unittest.TestCase):
 
-    def test_single_place(self):
+    def run_single_place(self, num_workers):
         scope = fluid.Scope()
         image = fluid.data(name='image',
                            shape=[None, IMAGE_SIZE],
@@ -192,7 +193,7 @@ class TestStaticDataLoaderReturnList(unittest.TestCase):
             dataset = RandomDataset(SAMPLE_NUM, CLASS_NUM)
             dataloader = DataLoader(dataset,
                                     feed_list=[image, label],
-                                    num_workers=0,
+                                    num_workers=num_workers,
                                     batch_size=BATCH_SIZE,
                                     drop_last=True,
                                     return_list=True)
@@ -203,7 +204,7 @@ class TestStaticDataLoaderReturnList(unittest.TestCase):
                 assert not isinstance(d[0], list)
                 assert not isinstance(d[1], list)
 
-    def test_multi_place(self):
+    def run_multi_place(self, num_workers):
         scope = fluid.Scope()
         image = fluid.data(name='image',
                            shape=[None, IMAGE_SIZE],
@@ -213,7 +214,7 @@ class TestStaticDataLoaderReturnList(unittest.TestCase):
             dataset = RandomDataset(SAMPLE_NUM, CLASS_NUM)
             dataloader = DataLoader(dataset,
                                     feed_list=[image, label],
-                                    num_workers=0,
+                                    num_workers=num_workers,
                                     batch_size=BATCH_SIZE,
                                     places=[fluid.CPUPlace()] * 2,
                                     drop_last=True,
@@ -225,6 +226,12 @@ class TestStaticDataLoaderReturnList(unittest.TestCase):
                 assert isinstance(d[0], list)
                 assert isinstance(d[1], list)
 
+    def test_main(self):
+        paddle.enable_static()
+        for num_workers in [0, 2]:
+            self.run_single_place(num_workers)
+            self.run_multi_place(num_workers)
+
 
 class RandomBatchedDataset(Dataset):