[Bug fix] Fix concurrency bug reported at issue#755 (dmlc#823)

* upd

* fig edgebatch edges

* add test

* trigger

* Update README.md for pytorch PinSage example.

Add noting that the PinSage model example under
example/pytorch/recommendation only work with Python 3.6+
as its dataset loader depends on stanfordnlp package
which work only with Python 3.6+.

* Provid a frame agnostic API to test nn modules on both CPU and CUDA side.

1. make dgl.nn.xxx frame agnostic
2. make test.backend include dgl.nn modules
3. modify test_edge_softmax of test/mxnet/test_nn.py and
    test/pytorch/test_nn.py work on both CPU and GPU

* Fix style

* Delete unused code

* Make agnostic test only related to tests/backend

1. clear all agnostic related code in dgl.nn
2. make test_graph_conv agnostic to cpu/gpu

* Fix code style

* fix

* doc

* Make all test code under tests.mxnet/pytorch.test_nn.py
work on both CPU and GPU.

* Fix syntex

* Remove rand

* Add TAGCN nn.module and example

* Now tagcn can run on CPU.

* Add unitest for TGConv

* Fix style

* For pubmed dataset, using --lr=0.005 can achieve better acc

* Fix style

* Fix some descriptions

* trigger

* Fix doc

* Add nn.TGConv and example

* Fix bug

* Update data in mxnet.tagcn test acc.

* Fix some comments and code

* delete useless code

* Fix namming

* Fix bug

* Fix bug

* Add test for mxnet TAGCov

* Add test code for mxnet TAGCov

* Update some docs

* Fix some code

* Update docs dgl.nn.mxnet

* Update weight init

* Fix

* reproduce the bug

* Fix concurrency bug reported at dmlc#755.
Also make test_shared_mem_store.py more deterministic.

* Update test_shared_mem_store.py

* Update dmlc/core

tests/distributed/test_shared_mem_store.py

@@ -9,7 +9,7 @@
 import time
 import numpy as np
 from numpy.testing import assert_array_equal
-from multiprocessing import Process, Manager
+from multiprocessing import Process, Manager, Condition, Value
 from scipy import sparse as spsp
 import backend as F
 import unittest
@@ -45,8 +45,6 @@ def create_graph_store(graph_name):
     return None
 
 def check_init_func(worker_id, graph_name, return_dict):
-    time.sleep(3)
-    print("worker starts")
     np.random.seed(0)
     csr = (spsp.random(num_nodes, num_nodes, density=0.1, format='csr') != 0).astype(np.int64)
 
@@ -69,6 +67,7 @@ def check_init_func(worker_id, graph_name, return_dict):
         g.init_edata('test4', (g.number_of_edges(), 10), 'float32')
         g._sync_barrier(60)
         check_array_shared_memory(g, worker_id, [g.nodes[:].data['test4'], g.edges[:].data['test4']])
+        g._sync_barrier(60)
 
         data = g.nodes[:].data['test4']
         g.set_n_repr({'test4': F.ones((1, 10)) * 10}, u=[0])
@@ -86,8 +85,7 @@ def check_init_func(worker_id, graph_name, return_dict):
         print(e, file=sys.stderr)
         traceback.print_exc()
 
-def server_func(num_workers, graph_name):
-    print("server starts")
+def server_func(num_workers, graph_name, server_init):
     np.random.seed(0)
     csr = (spsp.random(num_nodes, num_nodes, density=0.1, format='csr') != 0).astype(np.int64)
 
@@ -99,16 +97,21 @@ def server_func(num_workers, graph_name):
     efeat = np.arange(0, num_edges * 10).astype('float32').reshape((num_edges, 10))
     g.ndata['feat'] = F.tensor(nfeat)
     g.edata['feat'] = F.tensor(efeat)
+    server_init.value = 1
     g.run()
 
-@unittest.skip
 def test_init():
     manager = Manager()
     return_dict = manager.dict()
-    serv_p = Process(target=server_func, args=(2, 'test_graph1'))
+
+    # make server init before worker
+    server_init = Value('i', False)
+    serv_p = Process(target=server_func, args=(2, 'test_graph1', server_init))
+    serv_p.start()
+    while server_init.value == 0:
+      time.sleep(1)
     work_p1 = Process(target=check_init_func, args=(0, 'test_graph1', return_dict))
     work_p2 = Process(target=check_init_func, args=(1, 'test_graph1', return_dict))
-    serv_p.start()
     work_p1.start()
     work_p2.start()
     serv_p.join()
@@ -117,10 +120,7 @@ def test_init():
     for worker_id in return_dict.keys():
         assert return_dict[worker_id] == 0, "worker %d fails" % worker_id
 
-
 def check_compute_func(worker_id, graph_name, return_dict):
-    time.sleep(3)
-    print("worker starts")
     try:
         g = create_graph_store(graph_name)
         if g is None:
@@ -129,14 +129,14 @@ def check_compute_func(worker_id, graph_name, return_dict):
 
         g._sync_barrier(60)
         in_feats = g.nodes[0].data['feat'].shape[1]
-
         # Test update all.
         g.update_all(fn.copy_src(src='feat', out='m'), fn.sum(msg='m', out='preprocess'))
         adj = g.adjacency_matrix()
         tmp = F.spmm(adj, g.nodes[:].data['feat'])
         assert_almost_equal(F.asnumpy(g.nodes[:].data['preprocess']), F.asnumpy(tmp))
         g._sync_barrier(60)
         check_array_shared_memory(g, worker_id, [g.nodes[:].data['preprocess']])
+        g._sync_barrier(60)
 
         # Test apply nodes.
         data = g.nodes[:].data['feat']
@@ -167,15 +167,18 @@ def check_compute_func(worker_id, graph_name, return_dict):
         print(e, file=sys.stderr)
         traceback.print_exc()
 
-
-@unittest.skip
 def test_compute():
     manager = Manager()
     return_dict = manager.dict()
-    serv_p = Process(target=server_func, args=(2, 'test_graph3'))
+
+    # make server init before worker
+    server_init = Value('i', 0)
+    serv_p = Process(target=server_func, args=(2, 'test_graph3', server_init))
+    serv_p.start()
+    while server_init.value == 0:
+      time.sleep(1)
     work_p1 = Process(target=check_compute_func, args=(0, 'test_graph3', return_dict))
     work_p2 = Process(target=check_compute_func, args=(1, 'test_graph3', return_dict))
-    serv_p.start()
     work_p1.start()
     work_p2.start()
     serv_p.join()
@@ -185,8 +188,6 @@ def test_compute():
         assert return_dict[worker_id] == 0, "worker %d fails" % worker_id
 
 def check_sync_barrier(worker_id, graph_name, return_dict):
-    time.sleep(3)
-    print("worker starts")
     try:
         g = create_graph_store(graph_name)
         if g is None:
@@ -213,14 +214,18 @@ def check_sync_barrier(worker_id, graph_name, return_dict):
         print(e, file=sys.stderr)
         traceback.print_exc()
 
-@unittest.skip
 def test_sync_barrier():
     manager = Manager()
     return_dict = manager.dict()
-    serv_p = Process(target=server_func, args=(2, 'test_graph4'))
+
+    # make server init before worker
+    server_init = Value('i', 0)
+    serv_p = Process(target=server_func, args=(2, 'test_graph4', server_init))
+    serv_p.start()
+    while server_init.value == 0:
+      time.sleep(1)
     work_p1 = Process(target=check_sync_barrier, args=(0, 'test_graph4', return_dict))
     work_p2 = Process(target=check_sync_barrier, args=(1, 'test_graph4', return_dict))
-    serv_p.start()
     work_p1.start()
     work_p2.start()
     serv_p.join()
@@ -229,13 +234,28 @@ def test_sync_barrier():
     for worker_id in return_dict.keys():
         assert return_dict[worker_id] == 0, "worker %d fails" % worker_id
 
-def create_mem(gidx):
+def create_mem(gidx, cond_v, shared_v):
+    # serialize create_mem before check_mem
+    cond_v.acquire()
     gidx1 = gidx.copyto_shared_mem("in", "test_graph5")
     gidx2 = gidx.copyto_shared_mem("out", "test_graph6")
-    time.sleep(30)
+    shared_v.value = 1;
+    cond_v.notify()
+    cond_v.release()
+
+    # sync for exit
+    cond_v.acquire()
+    while shared_v.value == 1:
+      cond_v.wait()
+    cond_v.release()
+
+def check_mem(gidx, cond_v, shared_v):
+    # check_mem should run after create_mem
+    cond_v.acquire()
+    while shared_v.value == 0:
+      cond_v.wait()
+    cond_v.release()
 
-def check_mem(gidx):
-    time.sleep(10)
     gidx1 = dgl.graph_index.from_shared_mem_csr_matrix("test_graph5", gidx.number_of_nodes(),
                                                        gidx.number_of_edges(), "in", False)
     gidx2 = dgl.graph_index.from_shared_mem_csr_matrix("test_graph6", gidx.number_of_nodes(),
@@ -260,12 +280,20 @@ def check_mem(gidx):
     gidx1 = gidx1.copyto_shared_mem("in", "test_graph5")
     gidx2 = gidx2.copyto_shared_mem("out", "test_graph6")
 
-@unittest.skip
+    #sync for exit
+    cond_v.acquire()
+    shared_v.value = 0;
+    cond_v.notify()
+    cond_v.release()
+
 def test_copy_shared_mem():
     csr = (spsp.random(num_nodes, num_nodes, density=0.1, format='csr') != 0).astype(np.int64)
     gidx = dgl.graph_index.create_graph_index(csr, False, True)
-    p1 = Process(target=create_mem, args=(gidx,))
-    p2 = Process(target=check_mem, args=(gidx,))
+
+    cond_v = Condition()
+    shared_v = Value('i', 0)
+    p1 = Process(target=create_mem, args=(gidx, cond_v, shared_v))
+    p2 = Process(target=check_mem, args=(gidx, cond_v, shared_v))
     p1.start()
     p2.start()
     p1.join()