fix abort in mxnet shutdown, temporally use forked submodules

.gitmodules

@@ -1,9 +1,9 @@
 [submodule "3rdparty/mshadow"]
 	path = 3rdparty/mshadow
-	url = /~https://github.com/dmlc/mshadow.git
+	url = /~https://github.com/arcadiaphy/mshadow.git
 [submodule "3rdparty/dmlc-core"]
 	path = 3rdparty/dmlc-core
-	url = /~https://github.com/dmlc/dmlc-core.git
+	url = /~https://github.com/arcadiaphy/dmlc-core.git
 [submodule "3rdparty/ps-lite"]
 	path = 3rdparty/ps-lite
 	url = /~https://github.com/dmlc/ps-lite

include/mxnet/ndarray.h

@@ -848,14 +848,17 @@ class NDArray {
     // The shape of aux data. The default value for the shape depends on the type of storage.
     // If aux_shapes[i].Size() is zero, aux data i is empty.
     std::vector<TShape> aux_shapes;
+    /*! \brief Reference to the storage to ensure proper destruct order */
+    std::shared_ptr<Storage> storage_ref_;
 
     /*! \brief default cosntructor */
-    Chunk() : static_data(true), delay_alloc(false) {}
+    Chunk() : static_data(true), delay_alloc(false),
+              storage_ref_(Storage::_GetSharedRef()) {}
 
     /*! \brief construct a new chunk */
     Chunk(TShape shape, Context ctx_, bool delay_alloc_, int dtype)
-        : static_data(false), delay_alloc(true), ctx(ctx_) {
-      auto size = shape.Size();
+        : static_data(false), delay_alloc(true), ctx(ctx_),
+          storage_ref_(Storage::_GetSharedRef()) { auto size = shape.Size();
       storage_shape = shape;
       var = Engine::Get()->NewVariable();
       shandle.size = size * mshadow::mshadow_sizeof(dtype);
@@ -864,7 +867,8 @@ class NDArray {
     }
 
     Chunk(const TBlob &data, int dev_id)
-        : static_data(true), delay_alloc(false) {
+        : static_data(true), delay_alloc(false),
+          storage_ref_(Storage::_GetSharedRef()) {
       CHECK(storage_type == kDefaultStorage);
       var = Engine::Get()->NewVariable();
       if (data.dev_mask() == cpu::kDevMask) {
@@ -881,7 +885,8 @@ class NDArray {
     }
 
     Chunk(int shared_pid, int shared_id, const TShape& shape, int dtype)
-        : static_data(false), delay_alloc(false) {
+        : static_data(false), delay_alloc(false),
+          storage_ref_(Storage::_GetSharedRef()) {
       var = Engine::Get()->NewVariable();
       ctx = Context::CPUShared(0);
       shandle.size = shape.Size() * mshadow::mshadow_sizeof(dtype);
@@ -897,7 +902,7 @@ class NDArray {
           const std::vector<TShape> &aux_shapes_)
         : static_data(false), delay_alloc(delay_alloc_), storage_type(storage_type_),
           aux_types(aux_types_), ctx(ctx_), storage_shape(storage_shape_),
-          aux_shapes(aux_shapes_) {
+          aux_shapes(aux_shapes_), storage_ref_(Storage::_GetSharedRef()) {
       shandle.ctx = ctx;
       var = Engine::Get()->NewVariable();
       // aux_handles always reflect the correct number of aux data
@@ -914,7 +919,8 @@ class NDArray {
 
     Chunk(const NDArrayStorageType storage_type_, const TBlob &data,
           const std::vector<TBlob> &aux_data, int dev_id)
-        : static_data(true), delay_alloc(false), storage_type(storage_type_) {
+        : static_data(true), delay_alloc(false), storage_type(storage_type_),
+          storage_ref_(Storage::_GetSharedRef()) {
       using namespace mshadow;
       CHECK_NE(storage_type, kDefaultStorage);
       // init var